Open Research Online Charting a course to an emerging ﬁeld of ‘research engagement studies’: A conceptual metasynthesis

The past decades in the UK have witnessed renewed interest by policymakers, research funders and research institutions in the engagement of non-academic individuals, groups and organizations with research processes and products. There has been a broad consensus that better engagement leads to better impact, as well as significant learning around understanding engagement and improving practice. However, this sits in tension to a parallel trend in British higher education policy that reduces the field to a narrow definition of quantitatively measured impacts attributed to individual researchers, projects and institutions. In response, this article argues for the mobilization of an emerging field of ‘research engagement studies’ that brings together an extensive and diverse existing literature around understandings and experiences of engagement, and has the potential to contribute both strategically and conceptually to the broader impact debate. However, to inform this, some stocktaking is needed to trace the different traditions back to their conceptual roots and chart out a common set of themes, approaches and framings across the literature. In response, this article maps the literature by developing a genealogy of understandings of research engagement within five UK-based domains of policy and practice: higher education; science and technology; public policy (health, social care and education); international development; and community development. After identifying patterns and trends within and across these clusters, the article concludes by proposing a framework for comparing understandings of engagement, and uses this framework to highlight trends, gaps and ways forward for the emerging field.


Introduction
How do the public (or the public and third-sector practitioners who represent and serve the public) get involved with research? Should they have a say in the type of research that is funded? How can they participate in the production, communication and evaluation of research? What obstacles prevent them from accessing, adapting and using pre-existing research? And why is it in the interest of professional researchers and research institutions to support their engagement with research?
Over the past decades, the UK has witnessed renewed interest in research engagement. Research institutions and funders are recognizing that the impact of research in wider society can be improved by engaging users and mediators in research processes as well as with research outputs (HEFCE et al., 2011(HEFCE et al., , 2017Nurse, 2015; NESTA/Alliance for Useful Evidence, 2016). This has called into question the relevance of existing research systems, relationships, roles and products, with implications for what should count as 'researcher development' (for example, Enright and Facer, 2016;Weller, 2011Weller, , 2014Holliman and Warren, 2017). In response, there has been an explosion of studies and initiatives from fields as varied as higher education studies, science communication, cultural heritage, performance arts, museum studies, design, health and social care, education, international development, management, sustainability science, geography and political science. Through interrelated concepts such as public engagement, community-based research, research partnerships, coproduction, science communication, public understanding of science, citizen science, practitioner research, practice research, public scholarship, and knowledge transfer, or translation, or diffusion, or exchange or mobilization, these fields have developed theories, models and frameworks for understanding and improving engagement, as well as numerous examples of how engagement unfolds in practice and interacts through its different discourses with institutional policy.
These studies and initiatives reveal the many benefits of research engagement, summarized in a recent editorial in Research for All as substantive, normative and instrumental (Duncan and Oliver, 2017: 230): good engagement has the potential to improve the quality of research (including, as several authors point out, epistemological ways of knowing and ontological ways of being -see, for example, Oswald, 2016;Facer and Pahl, 2017;Hall and Tandon, 2017); to contribute to fairer and more equitable civil-society practitioners) that incorporates literature from a range of academic fields and sectors of policy and practice. After outlining the methodology of the review, it then goes on to map the evolution of understandings of research engagement in the context of five distinct but interrelated UK-based policy domains. The final section of the article draws on a comparison of the themes, approaches and framings identified across these domains to propose a framework for comparing understandings of engagement, and concludes by using the framework to highlight trends, gaps and ways forward for the emerging field of research engagement studies.

Methodology
The literature on research engagement spans numerous academic disciplines and fields of study, as well as sectors of policy and practice. It includes a multitude of terms, theoretical underpinnings, methodological approaches and empirical sites. Its authors are often academics, but also practitioners, policymakers, consultants and activistssometimes working in collaboration and sometimes alone in the contexts of their own organizations and professional sectors. It takes a range of forms from peer-reviewed articles to project reports, toolkits, manifestos, blogs and personal reflections. The term 'literature' implies written texts, however resources that communicate understandings and experiences of engagement have also taken more creative and embodied forms (see, for example, Back and Puwar, 2012;Facer and Pahl, 2017).
Charting a route through this vast terrain is a daunting task. There are many existing literature reviews that focus on different elements of research engagement (for example, Hanley and Vogel, 2012;Stahl et al., 2013;Munck, 2014;Watermeyer and Lewis, 2015;Oswald, 2016) and are framed by largely normative and pragmatic decisions about what to include or exclude, based on the expertise and interests of the reviewer. Other approaches are grounded in a more 'systematic' review methodology (for example, Davies et al., 2015;Shaw et al., 2014) and operate within clearly defined parameters, establishing strict criteria to search for and select the relevant sources. Still others (for example, Aniekwe et al., 2012;Facer et al., 2012;Fazey et al., 2013;Burchell, 2015;Davies et al., 2015;Shucksmith, 2016) have taken a more consultative approach, inviting experts from different fields to contribute resources, and convening participatory workshops to establish common frameworks and agendas.
The study that forms the basis of this article drew on elements of all three approaches to review and synthesize understandings of research engagement. While a comprehensive and systematic interrogation was not feasible (given the breadth, depth and diversity of the literature), the review attempted to provide a broad-brush account of the conceptual landscape that might constitute an emerging field of research engagement studies. The three key objectives were therefore: • to map the evolution (or genealogy) of different understandings of research engagement as they emerged within key UK-based domains of policy and practice • to identify the conceptual patterns and trends within and across these different domains • to draw on this analysis to develop an iterative framework for comparing understandings of engagement and identifying commonalities, tensions and gaps to inform a way forward for the study of research engagement.
These objectives translated into a methodological approach that was organized into four phases: (1) defining the scope of the review (identifying the conceptual literature and establishing key domains of policy and practice); (2) mapping the evolution of theories, policies and practice within these domains; (3) mapping key patterns and trends across the domains; and (4) developing an iterative framework for comparing understandings of engagement and charting a way forward for the field of research engagement studies.
Defining the scope of the review and identifying the literature The review began with a consultation of key informants (n=11) working on research engagement from within a range of disciplines, fields and sectors in the UK: higher education (two academics and one HE practitioner); arts and humanities (two academics); international development (one civil society practitioner and one policymaker); public health (one academic and one public sector practitioner); education (one academic) and science communication (one academic). As well as providing a series of narratives of the state of the art of research engagement in their area, the informants also recommended recent reviews published after 2008 (including syntheses of literature, but also historical overviews of specific sectors of practice). These 19 reviews (set out in Appendix I) were analysed to identify key terminology and conceptual/methodological framings, and to start to map out the (inter)disciplinary/ (cross)sectoral landscape of research engagement (see Figure 1), which represented a mixture of disciplinary traditions (for example, public sociology, critical geography and philosophy of education); fields of study (for example, museum studies, cultural heritage, development studies, and science and technology studies) and sectors of applied policy and practice (for example, health and social care, education, management, participatory democracy).

Figure 1: First iteration of the (inter)disciplinary/(cross)sectoral research landscape
This landscape helped to frame the review and provide prompts for literature searches. Initial spatial and temporal parameters were set to limit the review to a contemporary period (with resources primarily published between 2002 and 2017) and a focus on the UK (including both resources published by UK-based authors and resources focused on the UK as an empirical site). The review was an iterative process consisting of four search cycles, each contributing to a refinement of the (inter)disciplinary/(cross)sectoral research landscape, and each revision of the landscape contributing to the identification of further resources and terminology for searches. Studies were identified: first, through the 19 review studies set out in Appendix I; second, through recommendations from the advisers; third, through searches in key databases and indices; and fourth, through searches in key journals (for details of search strategies, see Appendix II). This resulted in the identification of 165 theories, models and frameworks of research engagement (see Appendix III). These were analysed according to authorship (academic, practitioner or academic-practitioner collaborations) terminology, theoretical underpinnings and methodological approaches.
Developing a map of evolutions of engagement within key UK-based policy domains The final refinement of the (inter)disciplinary/(cross)sectoral landscape resulted in the identification of five key domains that represented the five most substantial amalgamations of the literature as well as a series of relatively distinct domains of policies and practices. These domains were: higher education (incorporating elements of management studies); science and technology (incorporating elements of sustainability sciences); public policy (specifically health, social care and education); international development (incorporating elements of political science and thirdsector studies); and community development (incorporating elements of geography, arts/humanities and design) -see Figure 2. While these domains were chosen because they each constitute a key policy area as well as specific sites of practice and interdisciplinary fields of study, the significant overlaps between them should be noted. For instance, higher education is, of course, a component of public policy; international development incorporates elements of science and technology as both a field of study and policy sector; and community development interacts significantly with public policy and shares some influences with international development. However, these domains were all identified as contributing a distinct set of terminologies and approaches, which all evolved in different (though interrelated) contexts of practice. The review focused on developments in each domain from the early 2000s, while providing a broad historical background from the latter half of the twentieth century. Within each domain, an account of the recent evolution of understandings of research engagement was developed through the following stages: {1) identification of key historical trends in policy and practice; (2) identification of key terminology and approaches to research engagement within different fields of study; and (3) identification of key understandings of research engagement. Drafts of each account were shared with the key informants, and feedback was integrated into the final versions (see the section on evolution of understandings, pp. 191-202).

Developing a framework for comparing understandings of research engagement
The final phase of the review involved comparing the understandings of engagement emerging from the different domains of policy and practice to identify key patterns and trends within and across each domain. This analysis informed the development of an iterative framework to capture and compare the different ways of understanding research engagement (see the section on comparing understandings, pp. 202-8).

Evolution of understandings of research engagement within key policy domains
Since most understandings of research engagement combine an element of theory with an element of practice or policy, it is important to understand the theoretical or conceptual basis of how engagement is understood in the context of its evolution within specific sites of policy and practice. This section provides an overview and comparison of the evolution of different understandings of research engagement in the context of the five UK-based domains of practice, policy and thought: higher education; science and technology; public policy (health, social care and education); international development; and community development.

Higher education
There is a lengthy tradition of writing (particularly from the fields of critical sociology and philosophy of education) on the purpose of 'the university' and its relationship to 'society'. Some of this work has a spatial focus, exploring academia's relation to the 'knowledge society' (for example, Delanty, 2001) or a broader set of social ecologies (Barnett, 2009). Others have conceptualized the university's interrelated functions, either as a holistic model of 'scholarship' (for example, Boyer, 1990Boyer, , 1996, or as different spheres of academic function in tension with each other (for example, Burawoy, 2012), or through an epochal shift between Mode-1 and Mode-2 knowledge (for example, Gibbons et al., 1994) or the displacement of a 'culture of autonomy of science' by a 'culture of accountability' (Nowotny et al., 2001). More recently, a range of studies has explored the physicality and materiality of the university in relation to sociopolitics (for example, Cochrane and Williams, 2013;Temple, 2014). Other studies have focused more normatively on the university as a 'public good' (Marginson, 2007;Calhoun, 2011) and developed metaphors such as 'the civic university' (Goddard, 2009); 'the engaged university' (Watson et al., 2011); 'the open university' (Peters and Roberts, 2012); 'open scholarship' (Weller, 2011(Weller, , 2014; the 'engaged academic' (Cresswell and Spandler, 2013); and 'inclusive research' (Nind, 2014). Still others have focused on the public duty and/or value of different disciplines, for example, 'public sociology' (Burawoy, 2005), 'new public social science' (Brewer, 2013) or the 'public value of the humanities' (Bate, 2011;Belfiore and Upchurch, 2013).
Against this theoretical work, a parallel policy stream has developed the notions of 'research impact' and 'public engagement with research'. As the Research Assessment Exercise (RAE) evolved into the Research Excellence Framework (REF), a new measure of impact (defined as 'an effect on, change or benefit to the economy, society, culture, public policy or services, health, the environment or quality of life, beyond academia') was incorporated into the assessment (REF, 2011). In 2008, the National Co-ordinating Centre for Public Engagement (NCCPE) was established by the Higher Education Funding Council for England (HEFCE), and six further regional hubs were launched as 'Beacons for Public Engagement' to provide a coordinated approach to embedding a culture of public engagement (including recognition, reward and capacity-building) across the higher education sector. While the Beacon initiative ended in 2011, the NCCPE continued to receive core funding from the UK's research councils to embed the findings from the Beacons project within the wider sector, with initiatives such as the Catalysts for Public Engagement supporting this work. However, despite some conceptual work such as the AHRC-funded Researching the Engaged University Project, which brought together the academic and practice-oriented literature on 'university-public engagement' (see Facer et al., 2012), critics have identified a divorce between the 'alienating' vision of public engagement put forward by research funders and HE regulators, and academics' own histories of public engagement, which they frame through different terminologies and that link to broader academic traditions (see Watermeyer, 2011). Watermeyer and Lewis (2015) argue that, unlike the more coordinated evolution of similar discourses in other policy domains (for example, the public engagement movement in science and technology), the public engagement agenda in higher education was untheorized, suffered from lack of clarity in terms of its definition both as an activity and a topic for investigation, and subsumed the academic elements of the discourse under the administration of public engagement as a largely non-academic function. These tendencies are exacerbated by a further conflation of research engagement with a broader 'impact agenda', which might promote a more cynical instrumentalism, manifested as an attempt to be seen to be engaged rather than to be engaged for its own sake (Burchell et al., 2009;Pain et al., 2015).
However, the academic response to the 'impact agenda' has been varied, and this has also generated an additional range of understandings of engagement. While a first group of scholars have embraced the discourse of impact, focusing on establishing instrumentally 'what works' for better impact (for example, Van de Ven, 2007;Conway et al., 2009;Hughes and Kitson, 2012;Vostal and Robertson, 2012;Bastow et al., 2014), a second group, particularly from the arts and humanities (for example, Nussbaum, 2010;Bate, 2011;Docherty, 2015;Collini, 2012;Warner, 2015) have rallied to defend the university from new commodification and managerial control, arguing for preservation of existing structures and processes. According to these authors, a focus on the 'use' of research is an unhelpful starting point, as universities have an intrinsic 'cultural value' (Bate, 2011) or 'moral virtue' (Nussbaum, 2010). In this way, academic work plays a fundamental role in developing society's 'highest aspirations and ideals' (Collini, 2012: 86), so academic knowledge is a representation of a particular (higher-level) cultural vision for society and, therefore, necessarily detached from the context of its use. In contrast, a third group, of mainly social scientists, have argued that better engagement with society is essential for universities to critically engage with the neo-liberal trend in the impact agenda, critiquing the economic interpretation of impact and the commercialization of research (for example, Holmwood, 2010;Slaughter and Rhoades, 2004;Burawoy, 2012;Brewer, 2013). And, finally, a fourth, assorted, group have tried to generate alternative understandings of impact, for instance, by imagining 'feasible utopias' (Barnett, 2011) and reconceptualizing 'impact' from a feminist or collaborative perspective (for example, Pain, 2014;Pain et al., 2015;Facer and Pahl, 2017). Table 1 summarizes the different understandings of engagement within the higher education domain in relation to the evolution of policy and practice.

Science communication
In contrast with the higher education domain of policy and practice, the science and technology domain (and its academic manifestations as science communication -SC -and science and technology studies -STS) has experienced a relatively coherent evolution in conceptualizations of engagement from public understanding of science (PUS) to public engagement with science and technology (PEST). Despite the historical -and functional -necessity of strict segregation of laboratory-based science from society (to protect both the public from risk of contamination and materials from risk of contamination by publics (see Watermeyer and Lewis, 2015)) the communication of science beyond the university has a long history in the UK. In 1979, the journal Science Communication was established. In 1985, the importance of public understanding of science was flagged by the Royal Society through the Bodmer Report, which responded to a perceived crisis of public support for scientific and technological developments (see Burchell, 2015). This led to an agenda to educate the public and raise the national level of scientific literacy -an agenda substantiated by the establishment of the Committee on the Public Understanding of Science (COPUS) and the journal Public Understanding of Science in 1992 (Stilgoe et al., 2014). As many have noted, this agenda represents a deficit model of engagement, which perceives the public as uninformed and in need of education (for example, Brossard and Lewenstein, 2010). By 2000, however, the discourse had shifted to an emphasis on dialogue, as set out in the House of Lords Science and Technology Select Committee's Science and Society report (House of Lords, 2000), which called for two-way engagement with the public as part of the 'upstream' knowledge-production process -not just in relation to the translation of research outputs (Wilsdon and Willis, 2004). While deficit approaches include both public consultation and more educative information or awareness-raising initiatives, dialogic approaches are more collaborative and involve varying degrees of participation in the production, representation and communication of scientific knowledge. Both approaches, however, have implications for 'civic scientific literacy' (Miller, 1998) as a sort of capital to be transmitted (as with the deficit model) or as a prerequisite for participation (as with the dialogic model), with some engagement activities focused on outreach in schools (for example, Holliman and Davies, 2015). The dialogic model also interacts with some conceptualizations of citizen science, which range from public participation in data-gathering or crowdsourcing, interpretation of research and contributing to the design of research itself (for example, Bickerstaff et al., 2010;Haklay, 2013). Lewenstein (2005) also highlights additional types of engagement, which transcend the knowledge-policy divide. These include public engagement in deliberative processes around scientific policymaking and the engagement of research scientists in the democratic policy process. In this context, the Sciencewise-Expert Resource Centre (S-ERC) was established in 2007 as a resource to support the breadth of engagement activities in the science and society movementorganized via a 'public engagement triangle' into activities that 'transmit', 'receive' and 'collaborate' (Sciencewise, 2010).
Despite the range of engagement activities recorded under the PEST banner, critics have argued that this dialogical approach is often limited to specific scientific fields and issues, such as climate science and biomedical studies (Stilgoe et al., 2014). Others have problematized the singular conceptualization of 'the public' and 'experts', or 'society' and 'science', as discrete and dichotomized entities (for example, Maranta et al., 2003;Mahoney, 2012). In response, a smaller subset of the STS literature identifies an alternative approach to engagement based on a more contextual, relational and heterogeneous distribution of participation and expertise (for example, Collins and Evans, 2002;Jasanoff, 2004;Irwin, 2006;Chilvers and Kearnes, 2016) across 'deliberative systems' (Mansbridge et al., 2012) or 'ecologies' (Chilvers and Kearnes, 2016;Irwin and Horst, 2016). In this co-productionist or contextual approach, publics are understood as mediated, emergent, material and diverse collectives (rather than imagined aggregates of autonomous individuals who are external to participation), and participation is seen as a non-linear, multiply productive set of collective practices (as opposed to the traditional cause-and-effect model) (Chilvers and Kearnes, 2016). This implies that concepts such as 'participatory democracy', 'science' and even 'participation' itself cannot be taken as pre-given but are rather emergent affects of participatory assemblages. According to this conceptualization, then, engagement should not be seen as embodying a two-partner science-society interaction, but instead as involving a diverse range of groups who claim to speak for both science and society. Crucially, the constitution of these groups will change over time.
However, as with the higher education domain, the STS conceptualizations are also influenced by the impact agenda, as well as policy agendas beyond academia. Owen et al. (2012), for example, highlight a recent shift in terminology by the European Commission from 'science in society' to 'responsible research and innovation', exhibiting superficial political (and corporate) appeal, while contributing to more instrumentalist discourses of engagement. Table 2 summarizes the different understandings of engagement within the science and technology domain in relation to the evolution of policy and practice. While this third domain incorporates elements of the previous two (with overlaps between medicine and science policy, and interactions with the impact agenda of the higher education sector) it is also informed by several distinct traditions.
The first of these relates to patient and public involvement (PPI) in health and social care research. There is an extensive history from the 1970s of health activism driven by excluded groups in the UK running in parallel to global health social movements (see Brown and Zavestoski, 2004), which have influenced the integration of models based on 'shared decision making' into health policy and practice (see Ocloo and Matthews, 2016), although as Ocloo and Matthews (2016) conclude from their review of the sector, this involvement is still limited to consultation rather than collaboration, and involves a somewhat tokenistic subset of primarily less marginalized patients. In terms of patient involvement in research, a similar discourse led to the establishment of INVOLVE in 1996 as a national advisory group, in time housed in, and funded by, the National Institute for Health Research to support active public involvement in NHS, public health and social care research. Although an initial conceptual distinction was made between involvement at the levels of 'consultation', 'collaboration' and 'user control', this evolved into a more nuanced focus on overlapping approaches, each including multiple ways of involving members of the public in the identification, prioritization, design, implementation and dissemination of research (INVOLVE, 2012). From the field of education policy, a similar distinction is made by Rickinson et al. (2011), who identify five key approaches to working with 'research users': (1) creating feedback loops at various points in the research process, as well as in response to outputs; (2) university-led participatory research that integrates the perspective of 'users' to varying extents; (3) combining small-scale studies through the process of 'systematic review' or 'metasynthesis', which acknowledges 'grey literature' from the public or third sector as well as academic studies; (4) co-research for conceptual development or the design of studies; and (5) user-led research, which can be independent of universities.
Linked to the evolution of the policy, practice and understandings of patient, public or user involvement, a related trend focuses on practitioner research or practice research -the participation of public sector practitioners in research processes and practices (see Shaw and Lunt, 2017). This set of approaches has a strong Nordic influence from the fields of health, social care and social work, with an international position consolidated to some extent in 2009 through the Salisbury Statement. The statement contested the traditional idea that research informs practice, proposing instead 'that research also needs to be practice-minded in order to better study and develop knowledge which emerges directly from the complex practices themselves' (The Salisbury Statement, 2009: 4). However, within this broad definition, a distinction has been made between approaches based on partnerships between academics and practitioners, and approaches involving independent practitioner-led research (Shaw and Lunt, 2017). The latter set of these approaches links back to a tradition of 'teachers as researchers' (Stenhouse, 1975) within the UK's education sector, and a related tradition of teacher inquiry or practitioner inquiry from the United States (see Dana and Yendol-Hoppey, 2003). These approaches foreground reflexivity as a means of directing research towards oneself in order to understand and improve one's own professional practices, but have evolved to incorporate elements of action research approaches so that both learning and change are social and collaborative as well as personal (Mclaughlin et al., 2004). Nevertheless, these sets of approaches are framed at the level of the individual practitioner, and so contrast with two further approaches that conceptualize engagement at the level of the institution (in the case of education policy, the school). The first of these, located within the school improvement literature, focuses on the use of evidence to change school-based policy and practice, while the second conceives schools as knowledge creators in their own right (ibid.). All three of these perspectives have been incorporated into the British Educational Research Association (BERA)'s report Research and the Teaching Profession, which highlights the need for teachers to be able to access, interpret, critique and use education research (BERA, 2014). The report accordingly makes a strong case for schools and colleges to become 'research-rich environments', and for teachers to become 'research literate'. Working with 'research-engaged schools', Brown and Zhang (2016) highlight several factors as essential for evidence-based practice. These include: developing research capacity; creating a research culture; using research as part of an effective learning environment; and developing enabling structures, systems and resources. This body of policy-focused research tends to cast school leaders and governance systems as the primary points of engagement. In this way, it is notably different from more historical traditions of teacher inquiry, which foreground knowledge-production through and for reflexive practice.
A final evolving approach also focuses on the role of evidence in linking research to practice and links to an economic rationale for co-production (see Stephens et al., 2008). Emerging from the political climate of the late 1990s, and fuelled by the Blair Government's commitment to evidence-based policy, the establishment of the National Institute for Health and Care Excellence (NICE) in 1999 played a major role in synthesizing evidence for use across the sector. Shortly after, the Alliance for Useful Evidence was established by Nesta to engage both with the supply and demand side of evidence, and in the early 2010s ten additional What Works Centres were launched to continue this work within different sectors of public policy and different regions of the UK. Against this policy context, understandings of the relationship between research, policy and practice have evolved in three broad stages (or, as Best et al. (2008) term them, 'generations'): first, through linear models (based on dissemination, diffusion, knowledge transfer/translation, knowledge utilization and so on); second, through relationship models (based on knowledge exchange); and third, through systems models (based on knowledge integration). The recent systematic study by Davies et al. (2015) on 'knowledge mobilization' in the health, social care and education sectors identifies an increasing tendency towards the use of systems theory and complexity theory, and records a range of relational roles played by knowledge agencies in such systems. These include: knowledge production; brokering and intermediation; evidence advocacy; research into and in practice; fostering networks; and advancing knowledge mobilization. Understanding research-into-practice as a complex system, as opposed to a simplistic uni-linear process, also involves recognizing that knowledge mobilization is not just about research communication but also about 'access', 'uptake', 'adaptation' and 'utilization', and that these processes are not neutral but through processing, synthesizing, recycling, reinterpretation or adaptation transform, to varying degrees, the knowledge in question (Greenhalgh, 2010;Davies et al., 2015). Table 3 summarizes the different understandings of engagement within the public policy domain in relation to the evolution of policy and practice.

International development
This fourth policy domain interacts with an interdisciplinary field of study with roots both within and outside the university. As a field, development studies has sustained an interest in the intersection between knowledge-production, policy and practice, drawing on research from international policy and the NGO sector, as well as academic traditions of economics and political science (focused on structural analysis of the power relations between these sectors), and sociology, anthropology and geography (focused on ethnographic analyses of social practices of knowledge-production and use). Critics have also scrutinized the power relations inherent in contributions of different actors to knowledge about development (or the 'development discourse'; for example, Sachs, 1992;Escobar, 1995and, more recently, Moore, 2015. Complementing this analysis, a strong tradition of participatory research originating in the 1970s through the Latin American school of dependency theory and the work of the Brazilian educator Paulo Freire generated bottom-up or indigenous approaches to knowledgeproduction (for example, Chambers, 1997). While the participation movement began as a radical critique of the mainstream approach to objects of research (shifting power away from the expertise of the development consultant to the 'voice' of the poor), the approach -and particularly its manifestation in participatory rural appraisal (PRA) -was rapidly mainstreamed by international agencies, including the World Bank, and transformed into instrumental practice designed to gain access to communities and legitimize policies such as the structural adjustment programmes of the 1980s and 1990s (a precursor to austerity in the UK) and Poverty Reduction Strategy Papers of the early 2000s (see Munck, 2014). This led to a profound critique of participation as a 'new tyranny' (see Cooke and Kothari, 2001), masking power relationships and even exacerbating local power differentials. More recently, a related movement around 'cognitive justice' (Visvanathan, 1997(Visvanathan, , 2009Sousa Santos, 2007) and the role of 'southern theory' (Connell, 2007) has called for attention to diverse knowledges from the global South in a world where the production and dissemination of science and technology are still concentrated in the North (see also Hall and Tandon, 2017). This movement interacts with another participatory approach, termed community-based (participatory) research, which was particularly prominent in North America (see Israel et al., 1998;Wallerstein and Duran, 2003), although as Tandon and Hall (2014) point out, the engagement of research with communities has a longer, and more complex, history spanning Africa and Asia as well as Latin America. These critics argue that groups from the global South may view the notion of community (and its relationship to knowledge, nature and memory) quite differently to the dominant Eurocentric understandings that characterize Northern traditions around community outreach, civil engagement and service learning (see the following section).
However, this critical attention to inequitable global distributions of knowledge and power is by no means the only (or indeed the dominant) approach to research engagement in the international development policy domain. As with the public policy domain, studies on the role of evidence production and use in policy and practice have emerged from NGOs under pressure to satisfy donors and supporters by providing rigorous measures of success for their programmes (Eyben et al., 2015) in the face of the UK's Department for International Development's (DFID) 'payment by results' framework. In a sector deeply concerned with advocacy (in addition to improving understanding and organizational learning), it is vital for practitioners to be able to evaluate evidence (Hayman and Bartlett, 2013). Drawing on a similar body of literature to the public policy work around evidence-informed policy and practice, guidelines for assessing the quality of evidence have built on 'principles of credible research enquiry' (DFID, 2013: 10) and generated 'evidence principles' grounded in NGO values and types of intervention (BOND, 2013). In response, researchers working at the researchpolicy-practice interface have explored the implications for research capacity building across the sector (see Newman et al., 2012) and the role of NGO practitioners as researchers in their own right (see Hayman et al., 2016). This work has also fuelled a renewed interest in research partnerships between NGOs and universities (for example, Aniekwe et al., 2012;Hanley and Vogel, 2012), as well as critical analyses of how the different notions of evidence advanced by different stakeholders frame partnerships (for example, Cornish and Gillespie, 2009;Beardon and Newman, 2011;Shutt, 2009;Eyben et al., 2015). A recent focus on the role of 'evidence artefacts' in development policy and practice (see Eyben et al., 2015) has also introduced a new focus on materiality into understandings of knowledge-production -acknowledging the power of texts, templates and technologies to influence practice.
While studies on the politics of evidence for development have tended to focus on the evidence work of UK-based universities and international NGOs, a recent policy development in 2015 saw the launch of the Global Challenges Research Fund (GCRF) with a substantial investment of the British government's Overseas Development Assistance (ODA) budget into research funding. This has had a profound effect on the sector, with new incentives for UK-based higher education institutions to forge partnerships with researchers and civil society organizations from the global South, although critics have raised concerns about the allocation of ODA funds primarily to British universities as a type of 'tied-aid' (see ICAI, 2017). Table 4 summarizes the different understandings of engagement within the international development domain in relation to the evolution of policy and practice.

Community development
In this final domain, discourses including those identified in the previous domains (for example, around public engagement with research, community-based research and university-community partnerships) have evolved and intersected with other traditions of research and practice, as well as UK-specific policy shifts. O'Brien and Matthews (2016) provide a helpful account of these shifts in relation to (post-) urban regeneration, from the programme of regeneration by the New Labour Government in the late 1990s (specifically, the New Deal for Communities) to the Conservative Government's austerity agenda twenty years later (with the 'Big Society' programme of the Conservative-Liberal Democrat Coalition Government from 2010 and the Localism Act in 2011 -see also Allmendinger andHaughton, 2012 andMayo et al., 2013). This potted history helps to locate the renewed emphasis on communities as stakeholders in research. Against this backdrop, a series of interrelated approaches to research engagement can be traced.
The first of these relates to the lengthy history of research being conducted by individuals and groups outside of the university. In her extensive overview of nonacademic research practices, Finnegan (2005) lists, for example, seventeenth-century village astronomers, missionary meteorologists, Victorian amateur botanists, industrial investigators, local archaeologists, freelance family historians and internet bloggers. Some of these 'independent researchers' have strong activist identities, linking knowledge and advocacy through social movements around civil or disability rights, or in relation to environmental or social justice campaigns (for example, Morris, 1991). Others, such as 'hacker' and 'makers' communities, combine cooperative, creative and sometimes explicitly ideological practice with technical innovation (for example, Gaved and Mulholland, 2010) to develop new knowledge, as well as new objects and networks. Still others contribute to the development of formal research practice; for example, a particularly strong tradition in the UK has been that of oral history, people's history or public history, which grew prevalent in the 1960s and 1970s and intersected with the scholarly fields of folklore studies and labour history, later becoming contemporary community history studies (see Kean and Martin, 2013;Lloyd and Moore, 2015). The central aim of this approach was to capture eyewitness accounts of historical events or to understand the workings of communities through the experience of their members in a particular place and historical period. This gave non-academic researchers a way to actively participate in making history, either outside of universities or through 'shared authority' with academics (Frisch, 1990). In the UK, this authority was partly recognized by the legitimation of community-based oral historians through the formation of the Regional Network of Oral Historians in 1993. Non-academic research has also evolved in other fields, for example, in the visual and performance arts through use of the arts in community health research (for example, Ings et al., 2012;Macpherson et al., 2014) and practice-as-research approaches from the performance arts, which highlight embodied and reflexive ways of knowing alongside more conventional types of academic knowledge (see Nelson, 2013;Barrett and Bolt, 2007;Freeman, 2009).
While this first set of research practices emerges from outside of the university, a second set of approaches focuses explicitly on collaboration between communities and higher education institutions. These approaches interact with traditions around service learning, civic engagement and community outreach originating in the United States and Canada (see Bivens et al., 2015), but in the UK context tend to be grounded in traditions of participatory or action research (for example, Reason and Bradbury, 2001;Brydon-Miller et al., 2003), which foreground the interface between theory and practice as 'praxis ', or what Facer and Enright (2016), drawing on Whitehead, refer to as 'living knowledge'. While much of this work focuses on reconciling diverse knowledges, a body of more recent literature from the field of social movement studies (for example, Chesters, 2012;Derickson and Routledge, 2015;Choudry, 2015) highlights the power relations between academic knowledge and the theoretical knowledge produced by activists in communities outside of the university. Cox and Nilsen (2007), for instance, show how the academic social movement literature can perform three damaging functions: (1) exploiting activist theorizing (while claiming credit for itself); (2) suppressing activist theorizing (when it challenges the definition of 'the field' that the literature ultimately seeks to assert); and (3) stigmatizing activist theorizing as 'ideology' (rather than analysis grounded in practical experience). Others have raised the uncomfortable question of whether a contradiction exists between the aspirations of activist scholars for truly engaged research and their simultaneous participation in competitive and hierarchical academic practices (for example, Hale, 2008;James and Gordon, 2008;Pain, 2014).
A final development in participatory research (which links back to some of the work emerging from STS in the science and technology domain and the focus on material artefacts from the international development domain) relates to the participation of 'more-than-human' elements, including animals in research (see Reason, 2005;Bastian, 2013), with implications for understandings of participants, research methodology and ethics.
Returning to the impact agenda, a third set of approaches in the community development domain stem from the 'crisis in the humanities' (O'Brien and Matthews, 2016) as these disciplines struggled to account for their 'public contribution' or 'cultural value' (see Bate, 2011;Collini, 2012;Warner, 2015). This crisis sparked four key responses. The first took a similar form to the early work in science communication, with an increase in humanities scholars' interaction with the media, involvement in public lectures and debates, writing for lay audiences, participating in literary and arts festivals, and working with galleries and schools (see Levitt et al., 2010;Burchell, 2015). The second involved a drive to cultural innovation and entrepreneurialism (for economic impact) (see Bate, 2011;Hughes et al., 2011). The third saw an increase in the use of the humanities, but particularly the arts, in multi/cross/interdisciplinary efforts to improve the communication of science-based disciplines (see Barry and Born, 2013). The fourth, and probably most influential, of these responses was the large-scale programme funded by the Arts and Humanities Research Council (AHRC) called Connected Communities, which launched in 2010 and integrated a rich tapestry of arts and humanities-based approaches, including: public history; community-engaged arts; feminist, critical race theory and postcolonial traditions; civil and disability rights approaches; environmental practice engaging indigenous and 'non-human' knowledges; cultural, material and visual anthropology; cultural studies; patient engagement and responsible innovation; action research and participatory action research; communities of practice approaches; co-design; and open innovation, commons and crowd perspectives . The framing of engagement that has evolved through this programme has emerged as a collaborative process between community and university partners, which is 'sociallysituated', 'discursive', 'embodied', 'political and economic' and 'complex', and which generates a 'living knowledge' (ibid.: 23). A major influence of this programme has also been the recognition of 'community partners' as co-investigators in governmentfunded research.

Towards a framework for defining the field of research engagement studies
The understandings of research engagement that emerge from the five domains above have evolved in a similar historical and geopolitical context, but through different approaches to thought and practice. Within each domain there are contrasting (and at times conflicting) interpretations of elements of research engagement, including: the 'whos' of engagement (for example, 'users', 'stakeholders' and 'beneficiaries' or dialogue/collaboration between publics and scientists; universities and communities; practitioners and academics -although critics such as Maranta et al. (2003) and Mahony and Stephansen (2016) have addressed the limits of these imagined or aggregated identities, while others such as Reason (2005) and Bastian (2013) have called for the recognition of more-than-human animal or chemical or technological participants); the 'whys' of engagement (for example, individual, organizational and sector-wide motivations and incentives for engagement linked to practical, personal, conceptual and symbolic purposes -see Davies et al. 2015 -or substantive, normative and instrumental benefits -see Duncan and Oliver (2017); although, as Rotman et al. (2012) point out, these are likely to change over time); the 'wheres and whens' of engagement (for example, within, outside of, or at the borders of the university; in local or socially situated cultural contexts -for example, the notion of African community-based research proposed by Van Wyk and Higgs (2012), at different scales and against the unequal global distribution of the knowledge economy); and the 'hows' of engagement (for example, institutional policy and strategy; the logistical organization of engagement initiatives; the different engagement activities; and the dynamics of collaboration). In order to identify trends across these different understandings of research engagement (as well as gaps in the literature and ways forward for a field of research engagement studies) a heuristic framework is proposed based on three dimensions: (1) configuration of engagement; (2) locus of engagement within knowledge-to-action processes and systems; and (3) analytical lens. These dimensions are discussed in turn.

Configuration of engagement
The first dimension summarizes different understandings of how engagement is configured: (1) as activity; (2) as relationship; (3) as process; (4) as system; and (5) as affect.
Research engagement as activity: At its most basic, research engagement is understood as an activity, initiative or event. This understanding stems from the early science communication literature, whereby publics were invited to interact with research findings (for example, European 'science shops') and the arts and humanities literature involving festivals and exhibitions. Embedded in this understanding are a range of deficit approaches to engagement including 'transmission', 'dissemination', 'outreach', 'awareness-building' or 'communication'. However, activities can also include an element of consultation.
Research engagement as relationship: Probably the most common manifestation of engagement in the literature is the idea of a relationship between two parties (for example, universities and communities, scientists and publics, or academics and practitioners). This implies a more dialogic approach to engagement based on 'participation' or 'collaboration'. However, such relationships are never neutral and seldom equal. Questions around whose expertise counts, who is cast as 'author' or 'audience' of the research, and who has ownership of the research design, process and products can help to think through power relations. Facer and Enright (2016: 68-72) provide a helpful typology for the relationships between academic and community partners in collaborative research: (1) 'divide and conquer' approaches, in which there is a clear division of labour according to different sets of skills and expertise; (2) 'relational expertise', in which participants keep their own roles/identities but also try to understand (and even 'temporarily inhabit') others' perspectives; (3) 'remaking identities', in which participants actively take on others' identities and jointly construct new ways of knowing; and (4) 'colonisation and confusion', when the desire to disrupt hierarchies without a genuine sense of contexts and practices results in chaos and potential abandonment.
Research engagement as process: A third configuration of engagement manifests as the linear or cyclical processes of research or knowledge-to-action (for example, Graham et al., 2006). These understandings define a range of discrete stages for engagement to pass through, and contain underlying assumptions about the type of change that will be generated through the process. Barry and Born (2013) argue that within the impact agenda, this tends to be expressed either through the logic of accountability or the logic of innovation. However, the authors also identify a third logic (of ontology) that is capable of transforming the nature of knowledgeproduction itself.
Process models are also prevalent in the literature grounded in action research, which is based on a cyclical relationship between research, reflection and action. According to Murphy and Fafard (2012) this approach represents an 'emancipatory' type of knowledge as opposed to more 'instrumental' and 'hermeneutic' types.
And, finally, understandings of engagement as a process also form the basis of much of the practitioner-generated literature (particularly in the public policy and international development literature) and are represented by 'theories of change' frameworks that increasingly form part of organizational strategy (see Jackson, 2013).
Research engagement as system: In this fourth configuration of engagement, engagement unfolds through a more complex set of relationships between people, institutions, things, ideas and specific places. In the literature that is grounded in political science or geography, this tends to focus on the workings of the (global) political economy and its distributions of power and knowledge (for example, Connell, 2007;Sousa Santos, 2007;Choudry, 2015;Hall and Tandon, 2017).
Other approaches based on network analysis, systems thinking and complexity theory (for example, Burns, 2013) tend to work more on defining the different elements of systems within specific contexts. This approach also forms the basis of much of the literature on 'knowledge mobilization' (see Davies et al., 2015).
Research engagement as affect: A final configuration (particularly from the STS literature) conceives engagement as the emergent outcome of complex relational assemblages of social and material elements (for example, Irwin and Michael, 2003;Strathern, 2004;Jasanoff, 2004;Chilvers and Kearnes, 2016). This understanding implies that engagement work cannot always be planned from the outset, and that ideas about impact are likely to change over the course of the engagement. It also provides a helpful commentary on the emerging field of engagement research studies, which is in a constant state of becoming and can therefore never be fully defined as it continues to evolve in response to changes in practice and thought.
As such, affective approaches to engagement call for some imaginative rethinking of the philosophical 'conditions of possibility' of subjective identities and social practices (for example, Enright and Facer, 2016), institutional structures and processes (for example, Burawoy, 2017;Barnett, 2012Barnett, , 2016 and material artefacts (for example, Barry and Born, 2013;Chilvers and Kearnes, 2016).

Locus of engagement
This second dimension addresses the location of engagement within broader processes or systems of research-to-action. These include: (1) engagement in knowledge production; (2) engagement in research communication; (3) engagement in research utilization; (4) engagement in research governance; and (5) engagement in impact and learning.
Engagement in knowledge production (design and implementation): Most of the engagement literature locates engagement (at least to some extent) within knowledge production. However, the processes and practices of knowledge production are particularly foregrounded in the literature on collaboration (from the community development domain) and partnerships (from the international development domain). A smaller body of literature, which looks at knowledge-production by practitioners in sites outside of the university and runs across all five domains, also shares this focus.
These studies are all interested in the question of how research is designed and implemented, who participates and to what extent, how these processes are mediated by different types of methods, skills and artefacts, and what types of infrastructure or enabling environment is necessary to support engaged research production.

Engagement in research communication (representation and dissemination):
The literature on public engagement and science communication (from the higher education and science and technology domains) tends to focus on the interaction of the public with research, while the literature on public policy has traditionally focused on knowledge transfer, translation and diffusion, although more recent studies have adopted a more holistic focus on the knowledge-to-action process or cycle. Within the community development domain, studies grounded in participatory approaches have focused on the politics of representation of research, while studies into open research have focused on accessibility.
Engagement in research use (uptake, adaptation and utilization): The literature on evidence-based policy/practice (from the public policy and international development domains) concentrates primarily on knowledge utilization and the relationship between uptake and impact. More recent manifestations based on 'knowledge mobilization' again take a more systematic approach and acknowledge the interrelated processes of production/communication. This literature also includes a focus on the engagement capacity or research literacy of research users.
Engagement in research governance (agenda-setting and evaluation): A much more limited set of studies (primarily from the public policy domain) has focused on the participation of different groups in research governance processes, including funding-review colleges, criteria-setting panels, ethics committees and evaluative teams. Once again, this tends to form a component of the more systemic literature on knowledge mobilization. However, a far greater number of studies have observed that engagement practice is often curtailed by research funding policy and practice.
Engagement in impact and learning: Finally, some of the engagement literature either conflates engagement with impact, or explores the potential for engagement to offer an alternative conceptualization of impact (for example, Pain et al., 2015;Facer and Pahl, 2017). This literature responds to questions around how to evaluate and attribute impact. Linked to this is a broader interest in the relationship between engagement and learning. This is manifested across the literature in two key ways: first, through consideration of the particular skills or capacity required for effective engagement (with implications for researcher development work), and, second, through identification of learning as a key outcome of engagement. Particularly emerging from the participatory studies from community development (for example, Hart et al., 2013;Facer and Enright, 2016), these studies understand learning to be an essential part of the knowledge-to-action research cycle. Within this process, a key element is critical reflection, which might include discussion of any tensions, contradictions and power relations between the research participants.
Understanding engagement through identities: This first analytical lens draws on psychosocial traditions to produce 'embodied' understandings of engagement practices based on subjectivities, biographies and emotions. Much of the literature explores the motivations for engaging in research. Other studies focus on the diverse roles that different participants play. Several studies highlight the issue of representation, showing how the idea of distinct participant identities is largely imagined, although others argue that even if participants cannot be reduced to simplistic labels, research identities are still important to many, and especially those identities of a hybrid nature. Terms such as 'pracademic', 'researcher-activist', and 'scholar-activist' have gained prevalence in recent years, as researchers struggle to make sense of ambiguous and ambivalent subjectivities. Linked to this, a final body of studies focuses on affect in research engagement, including the 'emotional politics of collaborative research', with implications for how identities are formed and contested through research engagement (for example, Griffin et al., 2013).
A focus on identities as an analytical lens lends itself to research designs based on a narrative methodology. This approach invites individuals to represent their own experiences of engagement through interviews or journalling.
Understanding engagement through practices: This second analytical lens builds on the identities lens by exploring the social interactions around the processes and products of research engagement. Rooted in local or cultural contexts ('socially situated'), this lens tends to foreground the day-to-day experiences and interaction of specific groups or communities. Since studies using a 'social practice' lens have a particular interest in how people communicate, several also focus on how communicative interaction is mediated by artefacts; how people learn and develop 'expertise'; and how marginalized voices or knowledges might be 'empowered'. Studies using this lens are also interested in the dynamics of participation -in which power is a key component. For example, in the field of international development, studies have interrogated the politics of participation in partnerships.
A focus on practices as an analytical lens lends itself to research designs based on ethnographic approaches. This involves intensive fieldwork or participant-observation in one or more carefully defined sites. It also implies the need for careful reflexivity on the part of the researcher to disentangle the emic accounts of engagement from the etic description of those accounts (see Pike, 1967).
Understanding engagement though institutions: This third analytical lens focuses on the agendas, structures and processes of research-producing and/or using organizations and their effect on establishing enabling/constraining environments for research engagement. In the higher education domain, for example, many authors adopted an institutional lens to query the purpose of universities in the context of changing accountabilities and encroaching commodification and regulation. Other studies in the review explored how the purpose of research engagement was driven by the context of the political economy. While some studies focused pragmatically on how institutional infrastructure can be developed to support engagement, others recognized the symbolic role that research engagement can play. Related studies highlight the role of power to show how different modes of research are ordered by knowledge hierarchies, determining what counts as evidence.
A key component of this approach is the Foucauldian notion of discourse, whereby 'disciplinary boundaries, researcher identities … and definitions of research quality and validity are not neutral but contested, subject to change over time and can be understood as sites of struggle for symbolic and political power' (Facer and Enright, 2016: 23). In response, studies using this lens tend to draw on methodological traditions based on policy discourse analysis.
Understanding engagement through artefacts: This fourth analytical lens focuses on the material conditions of engagement as realized through specific technologies, tools and texts. Many studies from the review considered the nature of 'research products' produced through engaged research processes. While traditional academic research tends to prioritize outputs such as peer-reviewed written publications, a recurring theme in the literature is the need for more accessible products and the importance of 'open access' depositories and data archives. In collaborative research, representations of non-academic knowledge may well take alternative forms that privilege audio, visual or embodied modes of communication, or draw on the affordances of new media and technologies to produce multimodal ensembles. Authors writing about communitybased research have identified outputs as diverse as websites and blogs, documentary films, exhibitions, artwork and performances.
Artefacts are also a key aspect of much of the engagement literature grounded in science and technology studies, which focuses on engagement as a social-material assemblage of people, institutions, ideas and things, with some arguing that the very publics who are presumed to form the basis for public engagement are themselves materially constituted.
Specific methodologies associated with this lens include analysis based on human-computer interaction, web analytics and social-material semiotics.
Understanding engagement through accounts: This final analytical lens adopts a 'meta' philosophical stance to explore how conceptualizations of engagement are constructed. This approach tends to result in 'epochal' accounts of engagement or 'grand narratives', such as Gibbons et al.'s (1994) distinction between 'Mode-1' and 'Mode-2' knowledge, or the distinction between 'public understanding of science' and 'public engagement' as two models of science communication. However, this lens can also be used in conjunction with the identities, practices or institutional lenses to explore how individuals or groups make sense of their experiences of engagement.
A focus on accounts as an analytical lens lends itself to methods based on rhetorical analysis -such as the following section of this article, which compares the evolution of conceptualizations of engagement in different research domains. Figure 3 sets out the conceptual scope of the proposed field of research engagement studies as a Rubik's Cube based on the analysis above. While this might provide a tool to help locate the different understandings of engagement, most understandings incorporate multiple elements of the different dimensions, with understandings often shifting over the course of an activity or study. The metaphor of a puzzle is also potentially helpful in understanding an emerging field attempting to make sense of itself (with meaning -and pleasure -derived precisely from that sense-making).

Conclusion
This article has attempted to chart out a new field of research engagement studies by comparing a range of understandings that have evolved within and across five domains of policy and practice. By drawing on this review it has developed a framework to distinguish between the different core elements of engagement, as well as key crosscutting themes and the analytical lenses that define the different understandings. Across this framework, the following trends can be charted: (1) A move towards conceptualizations that are grounded in practice, as well as research approaches that are at least informed by non-academics, if not produced entirely in settings outside of academia. This increasing recognition of independent research practice beyond the university has also spawned an interest in the political economy of research, and the relative worth of academic versus non-academic approaches to research in the world today.
(2) A move towards more holistic, integrated conceptual approaches, which emphasize complex and dynamic configurations of people, things, organizations and ideas. These approaches also tend to take reflexive responsibility for the ontological affect of particular conceptualizations (or accounts) of engagement on the practice of engagement itself. This suggests the need for an ethics of engagement research as well as practice.
(3) A tendency towards approaches based on inter/trans/multi/cross-disciplinarity, which, as Burawoy (2012) suggests, runs the risk of undermining the rigour of the field, as well as potential for critique and, as Barry and Born (2013) point out, has the potential to undermine certain disciplines, subsuming them under others. (4) Increasing recognition of the ideologically motivated tensions and contradictions between different elements of the impact agenda in the UK, and higher education policy more broadly. This has simultaneously undermined the coherence and effectiveness of certain approaches to engagement, while creating opportunistic spaces for more radical practice. (5) Increasing attention to the issue of research capacity and the identification of new 'engagement literacies', both within and beyond the university. This issue is addressed in different (and at times conflicting) ways across the literature, and ranges from the incorporation of reflexive learning into engagement initiatives, to development of key skills for engagement, to the role of institutions in developing research capacity.
This review has also identified a number of gaps in the conceptual literature: (1) Despite the emphasis on collaborative or dialogic approaches to engagement, the literature is dominated by academic conceptualizations, while practitionergenerated conceptualizations are far less common. There is also likely to be a mismatch between research outputs, which are compatible with what we refer to as 'the literature', and alternative outputs that might include conceptual elements but are communicated through modes and media other than writing, for example, as artworks, films, workshops, exhibitions or performances (see, for example, resources described by Back and Puwar, 2012;Facer and Enright, 2016;Facer and Pahl, 2017). While collaborative initiatives have generated numerous academic accounts of these community-based representations, there is necessarily some translation involved, as they are re-contextualized into more conventional academic formats.
(2) Linked to the above, is analysis of the role of different types of 'engagement artefacts' in mediating (enabling or constraining) engagement practice. While the literature on open or digital scholarship is inherently optimistic about the democratic affordances of new technologies, there is need for a thorough appraisal of these in relation to more traditional research texts, technologies, tools and techniques. This also carries implications for the types of 'engagement literacies', which are, and should be, valued, and the types of enabling environments and infrastructures necessary to develop and sustain them. (3) Despite the conceptual shift towards 'systems' or 'ecologies' of engagement, there have been few attempts to trace engagement practices from research agenda setting and governance, to knowledge production, to communication, to use, and onwards to 'impact'. Such a study would add an invaluable empirical component to the conceptual literature. (4) Any conceptual analysis of research engagement must engage with its continuously evolving sites of production. Policy processes such as those surrounding Brexit and recent discourses around, for example, 'post-truth politics' will have a significant influence on how the relationship between knowledge and practice is understood and negotiated in the UK context. A field of study focused on understanding this relationship must be responsive to such geopolitical and sociocultural shifts.
Notes on the contributor Jude Fransman is a research fellow at the Open University's Institute of Educational Technology. Her interdisciplinary background spans education, international development, and science and technology studies, and she has worked in a research capacity for international organizations (including the OECD Development Centre, UNESCO and ActionAid international), as well as in academia. Her current research focuses on the politics of knowledge mobilization in response to global challenges and she coordinates an international network on Rethinking Research Partnerships: https://rethinkingresearchpartnerships.com • a collaborative relationship between academic and non-academic individuals, groups, institutions or sectors (for example, co-production or universitycommunity partnerships) • a research activity conducted by non-academic individuals, groups, organizations and sectors outside and independently of academia (for example, practice research, teacher inquiry or citizen science).
The strategies for these four searches were: • Given the breadth (and conceptual depth) of literature identified through these searches, a systematic classification and interrogation of the citations was not attempted. Rather, the searches were used to identify key terminology, as well as fields of study, subject areas, disciplines and sectors that contributed to the gradual refinement of the five UKbased policy domains and identification of key trends within those domains. This was substantiated though consultation with the key informant advisers.  Burns (2013) Model of factors influencing the shape and level of science-NGO collaboration

International development
Čada and Ptáčková (2012) Mode 3 knowledge production International development Carayannis and Campbell (2009) Theory of participation International development Chambers (1994Chambers ( , 1997 Southern theory

International development
Connell (2007) Participation as a new tyranny International development Cooke and Kothari (2001) Model/framework/theory  Stone and Pratt (1994) Collaborative perspectives and organizational relationships International development Sullivan and Skelcher (2002) Cognitive justice International development Visvanathan (1997Visvanathan ( , 2009 Community-based participatory research Community development Wallerstein and Duran (2003) Dialogic co-inquiry Finnegan (2005) The ontological politics of research in/on/with/by communities Community development Fransman (2017) Networked communities Community development Gaved and Mulholland (2010) A framework for institutionalizing community-university research partnerships Community development Hall et al. (2015) Framework for auditing, benchmarking and evaluating university-public engagement Community development Hart et al. (2009) Applying a COP approach to knowledge mobilization in community-university partnerships Community development Hart et al. (2013) Typology of university-community approaches to engagement Community development Holland and Ramaley (2008) Relational accountability in community-university partnerships Community development Kajner et al. (2011) Public history Community development Kean and Martin (2013) Sedimented histories

Community development
Lloyd and Moore (2015) Knowledge flows in interdisciplinary and community-based research Community development Lutz and Neis (2008) Community research for community development  (2012) Two traditions of community-based participatory research Community development Wallerstein and Duran (2003) Art as the production of knowledge Arts and humanities Barrett (2010) The public value of the humanities Arts and humanities Bate (2011) Beyond 'text methods' in co-produced research

Arts and humanities
Beebeejuan et al. (2014) Reframing the public value of the humanities Arts and humanities Belfiore and Upchurch (2013)