Everyday stories of impact : interpreting knowledge exchange in the contemporary university

Evidence & Policy • vol x • no x • x–x • © Policy Press 2017 • #EVPOL Print ISSN 1744 2648 • Online ISSN 1744 2656 • https://doi.org/10.1332/174426417X14982110094140 Accepted for publication 25 May 2017 • First published online 29 June 2017 This article is distributed under the terms of the Creative Commons Attribution 4.0 license (http://creativecommons.org/licenses/by/4.0/) which permits adaptation, alteration, reproduction and distribution without further permission provided the original work is attributed. The derivative works do not need to be licensed on the same terms.


Introduction
Within the UK, the advent of the measurement of research 'impact' in the 2012 UK Research Excellence Framework (REF) for measuring the research performance of universities, has led to a renewed preoccupation with the use of academic evidence in policymaking.Whereas traditionally this has been a focus of policy studies, the debates and research on how, and indeed whether, academic evidence should influence policy has been broadened across academic disciplines (Pain et al, 2011;Slater, 2012;Flinders, 2013;Smith and Stewart, 2016).In particular, academic disciplines such as the arts and humanities now find themselves expected to deliver 'impact' in the specific policy sense of 'an effect on, change or benefit to the economy, society, culture, public policy or services, health, the environment or quality of life, beyond academia' (REF, 2012).The evidence on evidence use by policymakers questions a lot of the assumptions behind the 'impact' model used in the REF exercise: the idea that one study could be directly instrumental in a rational policymaking model has been shown to be demonstrably unrealistic and also likely to be inappropriate (Weiss 1979;Nutley et al, 2007); even broader models of knowledge transfer or translation have been found wanting due to the one-way nature of communication suggested (Davies et al, 2008).Indeed, much of the research on evidence-informed policy -including the broader project that this paper comes from (Connelly et al, 2015) -suggests that developing close, collaborative working between policymakers is the best way to allow exchange of knowledge between academics and policymakers and for academics to become part of the complex networks of policymaking (Best and Holmes, 2010).
Summarising the literature on evidence translation, and the literature that has emerged in the UK alongside the REF impact agenda, this paper highlights that these mainly focus respectively on: the demand for evidence from policymakers; the boundary between academia and policymaking; and academic critiques of, or concerns about, the impact agenda.Building from this literature, this paper makes an empirical contribution by focusing on academic practices themselves -a noted gap in the literature to date (Oliver et al, 2014).Using evidence from a group of academic teams who were producing outputs that were specifically designed to impact on policymaking, the paper begins to answer the question: what are academics doing, or attempting to do, when they seek to deliver impact in the policymaking process?Taking an interpretive stance (Yanow, 2000;Yanow, 2003) we focus on what meanings these academics bring to the process and, further, what meanings they take from trying to deliver 'impact'.We understand our research participants and objects as 'situated "knowers" and situated "knowns"' (Yanow, 2003, 231).The research was co-produced with a civil servant with the UK Government, [Rutherfoord] a social researcher, who conducted the interviews and some analysis with [Matthews].This co-produced research process provided unusual empirical insights that are not prominent in existing research.
The data and analysis highlights three specific barriers to delivering impact that we argue are worthy of further attention: individual biography; disciplinary identity; and institutional pressures, including the REF itself.In presenting this data we highlight the challenges of negotiating relationships within policymaking environments, and the importance of these for understanding the needs of policymakers, communicating successfully with them and, ultimately, delivering 'impact' through academic research.

The Research Excellence Framework, 'impact', and evidence translation
There has been a long-standing concern with the use of academic research in policymaking.In recent years this concern has grown, particularly within the UK, due to the pressures resulting from the 2013 REF which for the first time judged university research on its socio-economic 'impact'.While this is of particular concern within the UK, we spend some time discussing it here for two reasons: firstly it has led to disciplines outwith applied social science or natural science attempting to get their knowledge used as evidence in the policymaking process; and secondly because it is important in framing our findings below.The measurement of socio-economic 'impact'within the REF has led to debate, and criticism, across academic disciplines emerging from the UK (Pain et al, 2011;Bannister and Hardill, 2013;Smith and Stewart, 2016).These range from the view that 'impact' is an imposition of a neoliberal accountability regime, narrowly focused on instrumentality which limits academic freedom, that must be resisted (Slater, 2012: in reply to Pain et al, 2011); 'debates concerning the "crisis" or "tragedy" of political science' (Flinders, 2013, 150); and within policy-focused research areas, such as health policy research, there is also concern that 'impact' will lead to a focus on research that is instrumentally useful for central government, stymying academic freedom and more innovative research (Smith and Stewart, 2016).Finally, in terms of individual academic careers, and wider structural inequalities within academia, there is a concern that it will benefit certain researchers more than others.As Newman (2011, 480) argues, '[n]otions of EBP [evidence-based policy], knowledge transfer and other contemporary discourses… serve to summon particular kinds of academic subject and privilege particular forms of knowledge'.
When we look at how socio-economic 'impact' is understood within processes such as the REF, the rational model of policymaking still clearly dominates thinking of how academic evidence might be used within policymaking (Bannister and Hardill, 2013;May and Perry 2013).The REF guidance states: Impact is 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… Impact includes, but is not limited to, an effect on, change or benefit to: … the activity, attitude, awareness, behaviour, capacity, opportunity, performance, policy, practice, process or understanding… of an audience, beneficiary, community, constituency, organisation or individuals.(REF, 2012, 26) This is a very linear, direct model, where a piece of evidence is seen to have clear and direct effects on societal outcomes, one of which may be policy.This adopts, or assumes, a quasi-rational policy cycle into which evidence from academics can just slot (Smith and Stewart, 2016).'Pathways to impact' then become akin to communication strategies, marketing findings to a presumed audience that is interested.This is still the case even though policy analysis has, for decades, demonstrated that policymaking does not happen in this rational way (Cairney, 2016); analogies of translation or transfer are ill thought-out and unrealistic (Davies et al, 2008); and the role of evidence in the policy process is complex and multifaceted (Weiss, 1979;Parsons, 2002;Nutley et al, 2007).
Research on evidence use by policymakers has predominantly elaborated the needs and practices of policymakers.At its simplest this focuses on differences and crossing boundaries: exhorting academics to 'do better', understand policymaking better (often ignoring substantial differences within this category), or 'talk human' (Flinders, 2013, 150).More complex accounts highlight the competing demands on policymakers -particularly the political context -and the way they judge and choose what evidence to use when making decisions.This is exemplified by realist accounts, such as the classic text on evaluation by Weiss (1997), or the accounts by Pawson et al of why policymakers need knowledge of what works, why and in what circumstances, rather than the types of evidence ordinarily produced by positivist, empiricist social science (Pawson and Tilley, 1997;Pawson et al, 2005).
The barriers and facilitators to getting research into policymaking have been summarised in a recent systematic review of the evidence (Oliver et al, 2014).In the section on research and researcher characteristics, this review found that: Researchers themselves were described as factors affecting uptake of their research.Having a good understanding of the policy process and the context surrounding policy priorities was supportive of research uptake… A barrier to uptake was identified where researchers were described as having different priorities from policymakers, with pressure to publish in peer-reviewed journals.Researchers were valued more when it was clear they were nonpartisan and producing unbiased results and provision of expert advice was also reported as helpful.(Oliver et al, 2014, 6) As Nutley et al (2007, 298) argue, postmodernist and interpretivist accounts take these insights further and move us 'away from understanding research use as primarily an individualised process, to seeing it as something that is socially and organisationally situated'.Such research blurs the boundaries between academia and policymaking, offering more fruitful analysis as to why, and how, evidence is used by policymakers.This includes understanding the complex, dialogic nature of how academic evidence is used in policymaking systems (Best and Holmes, 2010).This focus on the 'how' highlights how socio-cultural, organisational and individual biases and meanings affect how evidence is brought into policymaking.However, much of this work focuses on policymakers, and how they use evidence.For example, it suggests that they: carry class bias that prevents them defining a policy problem adequately (Matthews, 2012); struggle in chaotic situations to marshal the vast extent of evidence (Wilkinson, 2010); and are operating in policy networks with a range of actors translating evidence for them (Smith and Joyce, 2012).
Following this line of research, we follow Smith's more realistic line of argument: From the perspective of much of the policy sciences literature concerning policy change, the question becomes not, 'why is public health policy not evidence-based?' but 'why would we ever assume it could (or should) be?' (Smith, 2014, 562) In this paper we seek to take this further and ask, why would academics want their evidence used by policymakers?The insights of science and technology studies have provided nuance to broader debates about the types of knowledge created by academic research (Freeman, 2009;Smith and Joyce, 2012), and the variable take-up of these in different policy areas.However, a focus on academics as individuals with biographies and normative political commitments, and on academic practices, has been noted to be rare (Oliver et al, 2014).Exceptions are the reflections of particular academics who have taken on roles of 'policy entrepreneurs' or 'boundary spanners' (for example, Newman, 2011), or those debates within individual disciplines representing disciplinary reflexivity (for example, Flinders, 2013).One area of specific focus comes out of the reflection of knowledge-exchange professionals increasingly employed by universities (for example, Phipps and Morton, 2013) .While these accounts add richness and complexity to accounts of evidence use, we suggest that returning to the 'two communities' model (Caplan, 1979) might actually help us conceptualise the policymaking process more usefully by focusing us on one 'community' -academics themselves.This is particularly the case as there is little research on what academics actually do and their motivations when they try to get their evidence into policymaking: most research asks about perceptions, not practices, and most is interview-based.Less than ten per cent of the research Oliver et al (2014, 3) reviewed used observational methods, and while more practice-based work exists there is a consensus amongst its authors that not enough is known (Nutley et al, 2007;Oliver et al, 2014, 9).In the rest of this paper we present evidence from academics who were attempting to deliver 'impact' to help fill this gap in the evidence.

Method
The interviews for this research were carried out in 2014-15, as part of a wider project exploring practices of evidence use and communication on both sides of, and across, the academic / policymaking border.The project was undertaken collaboratively with the UK Government's Department for Communities and Local Government (DCLG), and an important aspect of this was to research academics from the perspective of policymakers.This is very unusual, and allowed novel insights into both the policymaking worlds and academics' worlds.The focus of the project was three policy reviews that had aimed to translate academic evidence from a major research programme, funded by the UK Arts and Humanities Research Council (AHRC), into policymaking on decentralisation and community governance.Two of the projects were led by scholars who would consider themselves social scientists.The way they chose to present their evidence reviews was largely in response to what civil servants in DCLG requested.The other project was led by a scholar from an arts and humanities background.The outputs of this project explicitly tried to represent the evidence in a different way and creatively disrupt norms of communication and policymaking (for a more detailed discussion, see Connelley et al, 2012).The research team -the authors of this paper -comprised four members of the policy briefing teams, as well as an academic, Matthews, who was not involved in the policy briefing projects, and a civil servant analyst from DCLG, Rutherfoord, who was in the team which was the principal audience for the briefings.
The 11 interviews that comprise the bulk of the data presented here were carried out by Rutherfoord and Matthews.The sample was all the academics involved in the policy reviews; one team member was an artist brought into one of the projects as a co-investigator.Apart from this artist, all our participants were in 'Russell Group' research-intensive universities in the UK.All but one of the original review team members were interviewed.One interview was carried out by one interviewer over Skype for convenience, the rest were carried out by both interviewers, in person.
The methodology of the research was interpretive -seeking to understand how participants understood the process of developing the policy reviews.This interpretive stance also framed the nature of the questions asked of participants.The interviews were semi-structured, covering the process of carrying out the reviews, participants' motivations for doing research that had an impact, and wider working environment, and were open-ended allowing participants to reflect on their activities, emotions and motivations.This produced a rich discussion between the three parties in the interview; indeed one interviewee commented at the end of the interview that they were not expecting such a deep and reflexive experience.The questions included one about the REF, as this had come out as of particular interest in the preparation for the interviews by participants and the interviewers.Given the interdisciplinary nature of the policy reviews, the interview also inquired about the interface between social science and arts and humanities disciplines in carrying out the reviews.Interviews were transcribed and shared among all six researchers, including those who had worked on the initial policy reviews.The transcripts were inductively coded in NVivo by Matthews, with analysis and input from Rutherfoord and the wider project team.Given the co-produced nature of the research project, this created a deliberative, reflexive team.The ethics of sharing data among the team were negotiated on an ongoing basis as part of the process of co-production (see Beebeejaun et al, 2015, for a discussion of ethics within contemporary co-produced research).Due to the ease with which quotes may be attributed to participants, they are reproduced sparingly in the empirical section to preserve anonymity.
The interview data was supplemented by further material.Firstly, observational data was collected by the research team during the course of the fieldwork, for example at meetings with policymakers and seminars, and meetings between the academic team and civil servants within DCLG.Secondly, the team members were asked to produce reflexive blog posts during the project: at the start asking them to reflect on producing the original briefs and then to reflect on where they felt they were going with this co-produced project (this helped develop the interview schedules); and then at the end of the project reflecting on how they think academic practices needed to change.Thirdly, through the co-produced methodology, the collective process of reflection in person and by email, and the production of 'tools' (discussed further below in the Conclusion), added a further richness and thickness to the data and analysis.
In analysing the data we used an interpretive approach to understand the practices and meanings of the interviewed academics as 'local knowledge' of policy actors (Yanow, 2000;Yanow, 2003).The interviewing approach provided rich material for this analysis.Rather than abstracting data out of the interview transcripts thematically through thick coding, the analysis was sensitive to the story being told by individuals, the contexts individuals found themselves in, and in turn created, through their actions.Narrative structure of stories reveals causal interpretations used by participants -that an event came after, and was linked to, a previous event (Hollway and Jefferson, 2000).In reproducing the data below, this desire to be sensitive to the data was in tension with the confidentiality of the participants, so a level of richness has had to be sacrificed.
In interpretive analysis, the aim is to move between micro levels of analysis, such as metaphor, myth and stories, and build these up to rich descriptions of meaningmaking systems -variously described as cultural domains (Coffey and Atkinson, 1996) or traditions (Bevir and Rhodes, 2006).Both cultural domains and traditions should be internally cohesive and relatively stable, yet because they are systems that actors use to make meaning of the world, they are also amenable to change.In this case, the quantity of data did not allow us to reach such theoretical saturation, so it would be inappropriate to consider whether we had identified stable domains or traditions, a point we return to in the conclusion.
As well as these abstractive approaches to data analysis, the co-produced methodology involved interpretation and analysis of the data.The three-way discussion in interview allowed for interrogation of specific points and a reinforcing discussion on diagnosis of problems and possible solutions.The reflexive end-of-interview discussion, when the recorder was switched off, furthered this, along with discussion between Matthews and Rutherfoord.In the rest of the paper we thus present three stories that come out of the thickness of this co-produced analysis that tie together the accounts of our participants and bring to the fore key aspects of academic experience and behaviour that were important in terms of knowledge translation or exchange type activities: individual biographies and motivations; academic disciplines; and institutional pressures, including the REF.We conclude by highlighting a solution identified by most participants, the need for dialogue between researchers and policymakers, and the relation of this to space, both physically and metaphorically.

Career biographies as meaning-making
Career biographies were a common feature in the interviews.The biographies were diverse, and participants used them to explain their current practices in a variety of different ways, most often without explicit prompting.These narratives were meaningmaking for our participants in the way that they helped explain why they wanted to do research that had an impact, relating this to a normative desire to change the world, and among five participants related to previous work outwith academia, in policymaking or related areas.They helped explain why our participants behaved in certain ways and, as will be explored in the subsequent section, placed them within an academic discipline and norms.
In terms of explaining a particular normative mission of our participants these short biographies created a causal story.For those participants who were not explicitly wanting to change policy, there was a common normative desire in their research to not just explain the world, but to apply that knowledge in some way to alleviate problems, or give marginal groups a voice in contexts where they would otherwise be ignored.For example, one of our participants described their ongoing work with a local deprived neighbourhood as: "trying to do something that for me is a long-term project" of representing the communities in different settings to help them.Another participant understood their current activities and research practices by relating them to working with gypsy-traveller groups in community development work over 20 years ago -the injustice they had witnessed then, and their experiences of trying to tackle them through research-based activism.Another older professor explained how they were "still driven by I want to make a change to society.I actually want a better society, I want some of those problems actually solved".
Individual biographies were also meaning-making for participants who had worked outwith academia -it was used to justify why they knew how things are done in other (predominantly policymaking) contexts outside of the academy.For example, one respondent explained: So I mean [name] City Council was an interesting place to pick, it was doing things differently.Before that I used to work in [local authority] in an R&I [research and innovation] unit which was one of the biggest R&I units outside London and it was just doing a job that local government should have been doing, it was fantastic.But it was innovative, it was your leading edge.So if you've been always in those circumstances you are very critical in a positive way you know, not critical 'oh it's bad' but you know, 'how could things be better?' This was part of quite a long career for this participant, but it remained important to them in terms of helping them better understand policymaking contexts.
These biographies were also important in relating how comfortable people felt 'outside' academia, or working in a policy environment: Before I was an academic I worked for a political party and I've done some think-tank work.
The interviewee went on to explain: So I suppose I've always been interested in kind of what goes on outside the academy and wanted to do research that was policy-relevant.And I suppose the sort of research I've done which has been about public services obviously lends itself very well to that.
Another participant used their career biography in terms of places of employment to describe their 'socialisation' into a particular position that was comfortable working with policymakers: The places that I've worked, so I was at [applied university research centre] and then at [local government-funded research centre] and [applied university department] and then at [applied university research centre], they're all places that really position themselves at a kind of nexus between policy, practice and academia.And that appealed to me and I think that… so that's just how I've been socialised.
For our participants these biographies thus explained their approach to their research, and importantly whether doing applied policy-relevant research, working with policymakers, and communicating in a policy environment were things they had experience in, or were socialised to expect to do as part of their role as an academic.

Academic disciplines and meaning-making and boundary-producing
Throughout our data, the importance of academic disciplines was a key theme.We see this in three ways: their centrality to biography; how comfortable, or not, participants felt in policy contexts; and the use of meaning and practices from specific disciplines in boundary-marking.
Firstly, and leading on from the previous section, academic discipline was often a key part of our participants' biographies, for example: I was actually lucky enough to be taught by [academic] and a lot of really amazing people at university which actually really affected how I saw the world.I mean he's really powerfully present in my work still.So my book … is on actually revisiting and thinking about [academic's] legacy.So that's me.
The closing "So that's me" is particularly telling -from this participant's experiences of being taught by a leading academic, we are meant to then fully understand everything about this person, the type of academic they are and what they do.Such important figures appeared in other participants' biographies and could be understood as academic 'parents', imbuing our participants with a way of being an academic, and importantly being an academic within their specific discipline.
Secondly, and important for our analysis here, this sense of disciplinary self led to how our participants understood themselves within policymaking environments and how comfortable they felt.For example, one meaning-making practice for our participants who had engaged with policy-focused research in the past was to understand the research process in a client-researcher relationship, often with the research client understood as a 'policy customer'.The ambiguities and tensions this produces for researchers, especially where the client is a government body, have been explored in other research (LSE GV314 Group, 2014).It should be noted that in our research some of our participants were in university research units that relied on consultancy income, but this notion of a client-researcher relationship was not confined to these people.
This affected how these participants went about doing their commissioned reviews.Some participants worked from the basis that there was a research client and they had to produce the research this client wanted.In this case, these researchers felt there was an ambiguity around who this 'client' was, as the policy reviews were funded by the AHRC to deliver impact through being used by the DCLG.For these scholars, this presented them with dual lines of accountability.The teams led by social scientists sought confirmation of what was needed from the client and were quite amazed that the AHRC were not in a position to offer this steer, and changed their approaches accordingly: So I think we made a decision on both that I didn't know what the AHRC really wanted but I thought I knew what you [DCLG] wanted.And I thought that I was better able to do that.This same participant also described in a blog post how the AHRC seem to lose interest: This was different to my experience of working on commissioned research where the 'client' is usually critically engaged… [they]… Seemed to lose interest once commissioned, particularly in the sense of building an ongoing policy relationship.
Because of this, these teams were comfortable reporting to, and engaged much more closely with, the DCLG.This was clear in the outputs produced -the teams produced reports in formats suggested by the DCLG: a slide pack (a substantial PowerPoint presentation intended to be read as a document) and a report informed by a 1-4-20 format and genre (a single-page summary, a four-page executive summary and a 20-page report).
This contrasts with the participants from the other team who had very little engagement with the civil servants who were to be the 'customer' for their knowledge.One member of this team reflected on this in terms of a lack of confidence to work with the DCLG: I think it's been really interesting working with people like [other team members] who are much more confident… I'm not as confident with [civil servants] though… I'm not confident with the DCLG.This team member was from a humanities background, so had no relevant experience from their research to date, or from wider disciplinary experience.
More broadly, our participants reflected on their own experience of different disciplinary practices and how some disciplines, for example urban studies and urban planning, social policy and public health, were more comfortable operating in a policy environment.In these disciplines doctoral supervisors or peers would interact with policymaking, and the practice was valued.Some participants noted how where they worked was different, and were not as comfortable positioning themselves as policyoriented.For example, one of our participants reflected that: there are institutes sort of similar to us… which will produce these nice sort of two-page, easily email-able, etc summaries of their research and when I suggested doing it here I think most people were broadly sympathetic as long as they themselves didn't have to write it.But you know, some of them who were more sort of professoriate, had a view that well journal articles only want original work, so we can't be publishing things in two-page summaries.
Another participant told us that "people didn't want to have to the word 'applied' in our vision of the School because they felt that that would demean the quality of their intellectual work".Doing 'impact' in these departments risked undermining perceived scholarly quality and reputation and represented a disciplinary difference for these participants who had greater policy experience.
The importance of discipline as a meaning-making tradition was particularly apparent when our participants were discussing the boundaries between disciplines, or how disciplines were seen by "outsiders".The evidence reviews being produced were, by their very nature, interdisciplinary: they were taking knowledge from the arts and humanities, loosely defined, and applying this in a policy context traditionally dominated by the social sciences.Because of this, the differences between disciplines emerged in the everyday work of carrying out the project, which is common in interdisciplinary working (Barry et al, 2008).
The following two quotes from participants exemplify this, particularly in the way that the participants felt their background was being stereotyped or misunderstood: I think the difficulty I had with working with [social scientists] was that it was very much 'art was fun'.
There'd be a whole lot of stuff about kind of stereotype of social scientists, which I found very irritating because I don't think of myself in those categories.
The broader literature on interdisciplinary working suggests that it is most successful when it is focused on particular issues and leads to epistemological change in the disciplines coming together (Barry et al, 2008).The two quotes from different team members presented here suggest this epistemological transformation did not fully occur.One interpretation could be that the epistemological difference between the disciplines was too great to ultimately bridge; indeed, in interview one team member used their hands to gesture such a bridge failing to be built, their fingers not quite touching as they brought their hands together.
Further, when interpersonal differences and the stresses of working to a tight timescale got too much, the practices of academic disciplines as traditions that were meaning-making were a comforting place to withdraw to, stopping the progress of interdisciplinary transformation in its tracks: The discussions kind of oscillated between really fabulous conversations and then [arts and humanities scholar] withdrawing into… arts and humanities bunker, from my perspective.So some of the time it felt very 'Yeah, we can really get to grips with this stuff ' and then suddenly it would stop being like that.
One participant used a "deliberate boxing metaphor" in their blog post to describe how this tension manifested itself in project working.Another participant described it as a "clash": "I think it was a disciplinary clash that looked like a personality clash.I think underneath it it's a philosophical clash that is quite deep".
In terms of the policy review produced, the tension and stress of nearing the final deadline meant the social scientists resorted to their traditional disciplinary practices: producing a report in the format that might be expected by a policymaker in a UK Government department.The arts and humanities wanted to disrupt communication with artistic outputs, particularly a short film.What is more important for us here, in understanding the efficacy of academics getting evidence into policymaking, is that it demonstrates how important academic disciplines were for our participants.When individual identity was threatened through tensions between individuals, project management, disagreements on communication styles and the relative weight given to different types of knowledge in outputs, individuals appeared to hunker down and use their academic discipline as a cloak.
A final example of a participant demonstrating this disciplinary 'otherness' is useful here.In a reflexive moment, this participant describes how they were described by another academic from a different discipline and the impact of this: Participant: And [another academic] asked me, he asked me at this workshop 'Where are you from in disciplinary terms?' and I said 'I'm from Politics' and he went 'Oh, sorry'.

Interviewer: (laughs)
Participant: And then said 'Politics is not very political' and it really hit home to me.And it made me realise that a lot of political science work is very dominated by like… empiricism… and is not political in the same way that social policy is.
In data such as this, we therefore see the importance of academic disciplines, as a set of norms and behaviours, as meaning-making for our participants.While the literature on evidence translation does recognise disciplinary differences in getting evidence into policymaking, this tends to focus on the nature of the evidence itself: for example the debates as to whether policymakers prefer the 'objective' knowledge of quantitative social science (Nutley et al, 2007).Alternatively it focuses on how different disciplines are effective, or not, in communicating their findings (Flinders, 2013).Interpreting academic disciplines instead reveals a different issue: that these disciplines are part of academics' deep meaning-making repertoire, and how they make sense of the world and operate in it.This means, in turn, that there are complex intersections of these meanings with working in a policymaking environment.

Institutional pressures
As discussed in the literature review, the debate about bringing evidence into policymaking is framed, particularly in the UK, in terms of the pressures on the contemporary university and issues of performance management (Shore, 2010;Slater, 2012).Here, we want to use the specific insights generated when a civil servant asked academics about their practices to understand institutional pressures in a novel way.We focus specifically on how our participants discussed the REF as this was a clear theme within the data; for the civil servant analysing this data, it was the least understood aspect of academic working conditions for an outsider.In total, across the interviews 'REF' was mentioned 78 times (although this does include mentions by the interviewers).It should also be noted that the interviews were carried out either side of when the final REF results were published, so the topic was pertinent.
As with most commentary and research from within academia in the UK (see, for example, Smith and Stewart, 2016), our participants were ambivalent about the REF as it impacted on their professional development, some more than others: Because I think a lot of it's bollocks.And I know it's important but I feel quite strongly they shouldn't drive our academic lives.I try and make sure that stuff I do is helpful but I'm not going to let the REF drive my academic life.
For our participants there was pressure on them to produce the four, "four-star" papers from research projects.One of our participants contrasted this to the "fun stuff ": "I also don't need bits of research that won't lead to REF-able publications… You can't just keep having fun, do you know what I mean?", by which they meant policy and practice-relevant work that was not fully recognised by their institution.
While we want to focus on the REF here as an institutional pressure that led to specific practices, it must also be acknowledged for many of our participants, the REF was part of a wider set of performance pressures.For example, for one participant, being awarded the policy review grant was linked to their desire to make a good impression and develop networks within a new university context: It's maybe a bit of a nervousness on my own part that I was going to you know, a Russell Group kind of university and it was quite a big step up… So yeah, it wasn't that there is a target to meet those awards, it was more a sense of I want to give an impression of what kind of colleague I am and what kind of contribution I'm able to kind of make.
However, the main institutional pressure on our participants derived from the REF was the tension between writing papers and taking the time to develop strong, trusting relationships with policymakers that would allow effective translation of evidence into policymaking to occur.The overwhelming power that the REF had over the choices and incentive structures of our participants was apparent.While performance management regimes, with organisational and individual targets and annual reviews, are now commonplace, particularly in government (Matthews, 2015), this was something quite different.The way the participants framed so much of their working lives in terms of the REF was striking -it would be discussed without prompting as something that affected academic practices, and narratives would often include the REF in the storyline.
It was also clearly apparent in the broader fieldwork for the project that the REF model of impact was inconsistent with the real world of multiple pressures on academics' time and policymakers' resources.Given the inherent time-lag in journal publication, and without access to academic journals, or the time to read and translate them, it was highly unlikely a policymaker would be able to access academic knowledge and evidence in the format privileged by the REF.Our data would suggest that it is not sufficient for academics to re-package, or translate, their evidence in ways that might be more suited to policymakers.Rather, there is a need for policymakers to understand the broader practices and meaning-making traditions of academics.

Dialogue in complex environments
The stories of impact presented above suggest a marked difference in the approach of the three teams: two were comfortable in policy environments, and sought dialogue with the policy 'customer' to tailor outputs so they could be readily used by policymakers.The other team, on their own admission, were working in the dark, as a member of the team commented on how they treated the end of the project presentation: We actually then behaved in a very traditional academic way and said here are the outputs from this project, go figure.And we didn't.There was no dialogue, there was no opening up the space.
The learning from this for this team was the need to open up dialogue to ensure greater understanding of the policy environment.In their blog post commenting on the end-of-project presentation, the same team member noted that the research in this paper "grew out of a conversation that day with [civil servant] to the effect that we needed to carry on the dialogue".The opportunity of one of this team being embedded within the DCLG (see Connelly et al, 2016) enabled this to be tested further, as they reflected in interview at a comment directed at Rutherfoord: [Civil servant] very graciously sat down and spent a bit of time with me looking at my version of my… film, which I'd already realised within ten minutes of being in your office, I realised that film was no use whatsoever to you but I couldn't articulate why.So I had that conversation with [civil servant] and [they] offered to sit down and kind of go 'That bit's useful, that's not' kind of thing, which was incredibly useful.But without that opportunity of sitting in your office I would have no clue on what to leave in and what to leave out as much as I did have just from sitting in there.
As discussed in the literature review, this suggests that Kaplan's idea of the 'two communities' is still a useful heuristic for understanding the divide between academia and academics, and policymakers.
However, rather than crossing the boundary, or learning a new language, as metaphors of communities and translation suggest, our research suggests that an ongoing engagement with the complex system of policymaking is required.Put simply, some of the academics we interviewed had greater experience, and thus learning, about this than others.As one team member commented in a blog post, for academics this requires: holding a position of learner, improving and using reflective practices, constructing / contributing to a space for honest dialogue.
The tensions in academics' careers and the pressures put on them, as discussed in the previous section, seem to mitigate against this.As one team member commented in a blog post: "In principle my department likes me to do this stuff, it doesn't make any time for it".
The co-produced aspect of this research demonstrated that this is not a one-way process and that policymakers can learn from this level of engagement, experience and learning.As another participant commented in interview: One would hope that that leads to a much deeper, closer engagement between policymakers, obviously we're talking primarily civil servants and in terms of setting the research agenda, monitoring and reviewing that agenda, having a dialogue, making sure that the results in the research gets sort of fed into government.
We now conclude the paper by exploring the possibility for co-production to further the embedding of academic evidence in policymaking.

Conclusion
This paper has sought to begin to answer the criticism from many scholars of evidence translation that we do not know what academics are doing when they carry out work to translate evidence into a policy environment, although there is ample evidence of academics' views and perceptions of what they have done.The 'two communities' approach has been a useful heuristic for focusing on academics' practices and knowledge.Taking an interpretive stance, we have understood academics as bringing specific experiences and meanings to knowledge translation activities and encounters.Further, through these activities, our participants took new meanings as they understood specific contexts better and further developed a cross-cultural competence necessary for good quality translation (Connelly et al, 2016).
We have highlighted three particular themes from our data that were important when working with policymakers.Firstly was personal biography, and in particular previous experiences of working in policy-focused environments.This gave this group of our participants an understanding of the motives and working practices of policymakers, increasing their comfort in these environments.Second, and related, was the importance of academic discipline.This was particularly an issue given the interdisciplinary nature of the policy reviews our participants had produced.It had deeper significance though -through their own admission some of the disciplines our participants would locate themselves in had much greater collective experience of working with policymakers which they could draw upon.Further, there was a clear role for disciplines in creating boundaries for the identities of our participants -they were this discipline, because they were different to another discipline that did things differently.Finally, specific pressures on our participants had a lot of meaning for them, in particular the pressures of the REF and the contradictory pressures to produce high-quality research outputs (primarily peer-reviewed journal articles), and the recognition that activities that might, ultimately, lead to evidence being fed into policymaking processes might not be highly valued.
The sample of participants in this study were self-selecting and cannot be assumed to be representative of their disciplines as a whole.The interviews were focused around the particular policy reviews produced in a particular context and time (in the run-up to the REF).There is likely to be much variability across academics, disciplines and policy areas.However, we believe the interpretive approach to understanding and analysing their practices and 'local knowledge' of trying to do impact, and influence policymaking, is a fruitful avenue for the study of evidence translation to follow.Supplementing this, the co-produced nature of this research also provided particular insights, which might not be readily forthcoming from a project conducted solely by academics.
Returning to the discussion of the REF that opened this paper: instrumentally, if we want to improve the efficacy of evidence translation, this research has specific implications.Most obviously, as mentioned, is the fact that it highlights the tension inherent within the REF between producing high-quality outputs and impact with those outputs. 1 Normatively, as our findings suggest, many academics do want to have a positive impact on policymaking, making outcomes more just, or improving the efficiency and efficacy of policy.Our research points to specific barriers in this regard as well -this normative drive behind providing evidence for making policy decisions and delivering policy is very different to the need for objectivity, particularly in the UK Civil Service, as enshrined in the Civil Service Code with its explicit commitment to impartiality (UK Civil Service, 2015).This does suggest that, in some ways, two very different communities do exist, and there is a need for cultural competence and understanding across the boundary.We have suggested practical ways to overcome these barriers elsewhere (Connelly et al, 2016).
In their systematic review of what works in evidence translation, Oliver et al (2014, 4) found that the 'the role of relationships, trust, and mutual respect' was extremely important for getting research findings into the policymaking arena.That this research -and this paper -were co-produced with a policy analyst in the UK Civil Service is further evidence to this effect.However, to develop such relationships means that academics must enter what Newman (2013, 525) referred to as 'contact zones'.It was clear from our data that some academics had more skill, experience, and knowledge to do this effectively.Our data also suggests that greater cross-cultural competence is needed.This needs to move beyond the obvious awareness that policymakers might need evidence repackaged in particular formats, such as 'triple-writing' (Flinders, 2013).Policymakers need to have greater awareness of the motivations, pressures and diversity of academics, and be more confident in exploring these with the academics they have, or want to have, contact with.Academics also need to have greater reflexivity.Part of this may be awareness and caution in entering a complex, political environment: Newman (2013, 525) highlights 'the ambiguities and dilemmas associated with creative labour in these perilous times: how the very words one speaks can rebound as they become taken up in government discourse'.The further reflection we suggest that might be required is a consideration of more basic issues: what are your motivations for trying to get your research findings into policymaking?Do they align with the policymakers you are wanting to work with?Have you discussed this with them?While this may not produce a panacea to make the process of translating evidence into policymaking a more fruitful process, the cross-cultural competence this may engender might make these interactions slightly less haphazard.