Credibility in Policy Expertise: The Function of Boundaries Between Research and Policy

As science becomes an increasingly crucial resource for addressing complex challenges in society, extensive demands are placed upon the researchers who produce it. Creating valuable expert knowledge that intervenes in policy or practice requires knowledge brokers to facilitate interactions at the boundary between research and policy. Yet, existing research lacks a compelling account of the ways in which brokerage is performed to gain credibility. Drawing on mixed-method analysis of 12 policy research settings, I outline a novel set of strategies for attaining symbolic power, whereby policy experts position themselves and others via conceptual distances drawn between the “world of ideas” and the “world of policy and practice.” Disciplinary distance works to situate research as either disciplinary or undisciplinary, epistemic distance creates a boundary between complex specialist research and direct digestible outputs, temporal distance represents the separation of slow rigorous research and agile responsive analysis, and economic distance situates research as either pure and intrinsic or marketable and fundable. I develop a theoretical account that unpacks the boundaries between research communities and shows how these boundaries permit policy research actors to achieve various strategic aims.

decisions, substantially underestimate the role of politics in the science-policy interface (Jasanoff, 1990). Accordingly, there has been a rapid growth of think tanks and public policy institutes that take on a brokerage role to make scholarly knowledge useful to decision makers (Stehr & Ruser, 2017). Historically, science and politics were seen as divergent social worlds or subsystems, with differing conventions, actors, and cultures (Merton, 1949;Montpetit, 2011). Recent literature, however, demonstrates the multitude of movement and action that takes place at the sciencepolicy boundary. Knowledge utilization studies (Gano, Crowley, & Guston, 2006), advocacy coalition framework research (Weible, Sabatier, & McQueen, 2009), and policy network analysis (Robins, Lewis, & Wang, 2012) show how actors and institutions can successfully operate in both social worlds. It is increasingly clear that these worlds overlap, with a permeable border between research and policy (Newman, 2014;Smith, 2013). However, further development of the theoretical role of brokerage is required to adequately account for the intersection of different social arenas. I offer an approach that demonstrates how the boundaries between research and policy are established in language and practice. To varying degrees, policy researchers are accomplished in both research and policy, and thus rely upon these distinct categories in the way they position themselves in the pursuit of credibility. A critical task is to examine how policy experts construct and perform these boundaries.

Brokers
One key way the policy studies literature has dealt with the movement and translation of policy knowledge is with reference to the "broker" (Sverrisson, 2016). Early theory on brokerage was developed by Burt (1992), who pioneered a network approach that emphasizes the relations between diverse actors rather than individual characteristics or behaviors. Brokering has subsequently been observed in a range of settings (e.g., Johri, 2008;Vogel & Kaghan, 2001), and described using a range of terms, including "boundary spanners" (e.g., Valente & Fujimoto, 2010) and "mediators" (Di Marco, Taylor, & Alin, 2010). In particular, the concept of "knowledge broker" has been widely applied to those working in think tanks, universities, or other research contexts, to describe a means of brokering knowledge across boundaries by acting as intermediaries between academia and wider society (Osborne, 2004;Wachelder, 2003). Research has primarily focused on two elements of the broker: the communities or coalitions within which they are enmeshed (Emirbayer & Goodwin, 1994;Lave & Wenger, 1991) and the benefits they provide and risks they face (Bielak, Campbell, Pope, Schaefer, & Shaxson, 2008).
Current conceptions of brokerage emphasize the broker's role in bringing together diverse groups and individuals and facilitating the production, dissemination, and the use of knowledge within professional networks (Lomas, 1997;Plehwe, Riese, Miller, & Bührmann, 2012). Specific activities include liaising with stakeholders to develop research agendas, collaborating with external organizations, pursuing avenues of funding, communicating findings through diverse modes of output, and formal and informal training for other bodies (Cooper, 2014;Rickinson, Sebba, & Edwards, 2011). Within applied disciplines (such as health and education), the broker concept has been used to describe public and policy-oriented intellectual engagement (Cooper, 2014). In this context, nonexpert voices are seen to encroach on traditional disciplinary experts, creating greater accountability and forcing expertise into a public-focused, interdisciplinary, and socially useful model. Throughout this literature, however, there remains a tendency to view the connection between research and policy as a process whereby research findings move from the research world to the policy world, where they impact decisions and actions.
Research on brokerage tends to rely on specific characteristics or roles that are prescribed by an actor's particular context. For example, Pielke (2007) provides a theoretical framework of four idealized roles for scientists that depend on different conceptions of science (i.e., linear or stakeholder) and society (i.e., scientists as advocates or scientists as resources). Key here is the archetype of the "honest broker of policy alternatives," which represents "an effort to expand (or at least clarify) the scope of choice for decision making in a way that allows for the decision maker to reduce choice based on his or her own preferences and values" (Pielke, 2007, pp. 2-3). Yet, recent work has highlighted the relational nature of knowledge transfer. Bocher and Krott (2016) provide a "research, integration and utilisation" model of scientific knowledge transfer that offers a crucial link between science-based information and the resources of political and practical actors. The model holds that cutting-edge, independent research is transferred into effective, practical application by way of an active, bidirectional selection process of the research results that are relevant in practice. This line of work reorients us toward the flexible and ever-changing elements of interpersonal interactions that occur at the science-policy boundary. Thus, attention must be directed to how actors actively position and reposition themselves and others through a coherent strategic process, rather than merely acting out a prescribed role. There is a need for research that explores the differences between research and policy by considering the way these communities are constructed by research actors themselves.
Despite the recognition of the importance of the intersection between research and policy, studies on the policy process do not adequately capture the boundary position between science, politics, and other professions that policy experts occupy. Collective professional rhetoric is a critical feature of the systems that link specialized knowledge to practice (Latour, 1987), yet the forms and functions of this type of positioning have yet to be examined. Despite the popularity of brokerage models, therefore, existing research lacks a compelling account of how this type of role is performed in language and action in the pursuit of credibility. By investigating the strategies of a range of organizational types, I elaborate the ways in which brokering strategies differ across research contexts within an applied area of research oriented to politics, public policy, and practice. Specifically, I seek to unpack the brokerage role, by considering how conceptual differences between the "world of ideas" and the "world of policy and practice" are constructed and utilized by policy research actors.

Boundary Locations
There is a recent body of work that takes the production of applied or policy knowledge as a dynamic process that is often created in the space between more established or mature fields (Eyal, 2011). In this conception, policy research is not produced in contained and self-referential communities. Rather, it occurs within a boundary location made up of diverse actors who collaborate, struggle, and converge to create intellectual interventions. These varied sites, with a diverse range of expectations and requirements, destabilize the notion of bounded, coherent, and exclusive professions or disciplines inhabited by specified policy experts (Eyal, 2006). Yet, these sites also cannot be described as creating novel, separate, and fixed sites of expertise (Stampnitzky, 2011). Rather, they are liminal spaces at the nexus of a variety of professions, disciplines, and organizations with varied aims and strategies (Eyal, 2006). These boundary locations are notable because they allow members of a bounded profession or community to engage in activities outside their presumed roles (e.g., Greenwood, Suddaby, & Hinings, 2002). In this article, I consider the ways individuals and institutions rely on established disciplines, professions, and contexts to situate their own knowledge.
Within this "space between fields" (Eyal, 2011), research actors acquire meaning through a process of boundary-work. This boundary-work consists of symbolic relations enacted through the language and practices of those who have a stake in defining what research is, and who work to establish it in the social world (Gieryn, 1983;Medvetz, 2012b). In his work on think tanks, Medvetz (2012b) shows that boundary-work initially appears to operate simply through an elaborate process of differentiation. For example, by pointing to core funding and peer review, a university department can demonstrate it is not a think tank or government agency. However, from each differentiation arises concomitant affiliation. For instance, although university departments are held to operate in the detached pursuit of knowledge, this is no longer the case in the age of demonstrable impact. Thus, in addition to showing that it is not a think tank or government agency, a university department must now also demonstrate that it is not "useless," and therefore must increasingly introduce just enough political clout or business-like efficiency into its scholarly practice.
Given the uneven nature of these tensions, some research actors have to work harder to establish their location in the space between fields. For example, when establishing their "independence," university researchers have examples of academic practice close at hand, and fewer "opposing" practices (e.g., fundraising, politicking) than think tank or government researchers, but still often must position their intellectual labor as appropriately scholarly, marketable, or useful. In short, there is an ever-present danger in going too far in the process of affiliation (e.g., too political and one ceases to be an autonomous academic, too academic and one ceases to be valuable in practical terms). Thus, the requirements for gaining credibility in policy research shift from moment-to-moment as the process of affiliation and differentiation progresses, always working to maintain balance appropriate to the organizational identity or structural proximity to an established field. This process of affiliation and disaffiliation involves symbolic boundaries or "conceptual distinctions" (Lamont & Molnar, 2002, p. 168). Researchers seek to establish the dominance of their discipline, profession, or organizational type. At the same time, the divisions between disciplines (e.g., economists and others) and professions (e.g., consultants and academics) are seen to be "natural" divides that must be overcome in the pursuit of sound policy outcomes. In the making of policy knowledge, therefore, the boundary-work that individuals and institutions engage in involves both boundaries-as-exclusion and boundaries-as-natural-divisions (Riesch, 2010). The nature of this boundary-work depends on the positioning of an organization within the space between fields. Drawing on Baert and Shipman's (2012, p. 197) conceptualization of "epistemic distance" between expert and lay consumers of knowledge, and epistemic and temporal differences between think tanks and academics (Tchilingirian, 2015), I argue that the concept of "distance" is useful here.
I argue that within policy research organizations, a key overarching tension is a claim to academic capital (i.e., symbolic resources) and a claim to "policy and practice" capitals. In considering three research contexts-universities, think tanks, and government agencies-I examine how researchers situate their work via conceptual distances drawn between the "world of ideas" and the "world of policy and practice." In doing so, I unpack the concept of the knowledge broker. I seek to demonstrate how researchers and institutions negotiate symbolic power and credibility through ongoing self-positioning. I examine how research actors position themselves as distinct or similar to specific disciplines, professions, and contexts. For example, actors may differentiate their intellectual work with reference to either economics, sociology, or agriculture; and/or as a political operative, academic, or consultant; and/or as affiliated with a university, think tank, NGO, or government agency. I elaborate four distances, and demonstrate how they are established and employed.

Methodology
This article is based on a study of the boundary locations of policy knowledge and expertise between established disciplines, professions, and fields. Twelve policy research organizations, focusing on international development, were selected in the United Kingdom, Australia, and the United States (see Tables A1-A3 in the Appendix). In order to explore international development as a transnational field, the analysis was extended beyond a single country. The three chosen countries contain similarities that make features and structures of knowledge production recognizable across contexts, but that each represent different organization and characteristics of fields. The cases were chosen because they have established policy-relevant research programs; that is, those that were actively engaged in contesting and producing knowledge for development research, practice, and policy (i.e., excluding those who do not specifically publish research). The four university contexts included two research leaders and two with alternative outlooks. Staff numbers ranged from 8 to 65, and funding was primarily from government bodies, with additional funds from charities, research councils, and the private sector. The four think tank contexts included two with university affiliations and two with alternative models. Staff numbers ranged from 49 to 235, and funding took the form of research grants; contributions from philanthropic foundations, individuals, and governments; sale of knowledge services, teaching, trading; and membership fees. The four government agencies included two bilateral agencies as well as two dominant multilateral agencies. Exact staff numbers were unable to be attained, given that researcher lists were not published, and funding for these four cases came from internal program budgets.

Data Analysis
I conducted directed content analysis of interview data and institutional materials. Information was collected about many different aspects of the institutions via interviews with researchers and collation of documents and websites that gave a sense of institutional character. Attending to intellectual outputs permits an examination of the ways boundaries are maintained and transcended (Lamont & Molnár, 2002). Intellectual products are transient interventions within and between worlds, which create shared meaning among relevant actors without restructuring established fields. As such, analysis of outputs permits understanding of the processes, strategies, and outcomes of institutions and individuals. However, an exclusive focus on the intellectual products of organizations misses a crucial element. Boundaries and spaces between fields operate through active engagement by actors (Bourdieu, 1993). It therefore becomes important to allow researchers to elucidate what they consider meaningful, by attending to the flexible elements involved in the active positioning and repositioning of selves and others through discursive engagement (Moghaddam, 1998). Thus, theoretical attention to the ways identities are enacted through language is warranted. Thus, the changing context of intellectual labor as constructed by those working in boundary locations must be investigated alongside the outputs that are produced.

Data
The findings are based on interviews with 75 individuals, who were involved in the knowledge production process. Suitable participants were identified through university and institute websites. Participants were involved in producing and disseminating policy documents, academic articles, and other outputs in the area of international development. The sample list contained a mix of seniority and an even spread of genders. The sessions lasted between approximately 45 minutes and 1 hour, at participants' workplace or in a place convenient to them. The study was introduced as an examination of knowledge production in international development, and discussion was guided by an interview schedule that focused on characterizing the field, research practice and processes, structural features, and outcomes and impact. Consent was obtained in writing from each participant. Permission was gained to include the names of each of the case institutions. In order to ensure anonymity, participants are identified by pseudonyms that are linked to the category of organization (e.g., "a university researcher") rather than the named institution itself.
The findings are also based on a detailed analysis of documents, which entailed systematically reviewing and evaluating printed and electronic materials. Documents were taken to be naturally occurring "social facts," produced, disseminated, and used in socially organized ways (Atkinson, Coffey, Delamont, Lofland, & Lofland, 2001). The process involved examining both what is included and excluded by the texts, in addition to the imagery and dichotomies drawn upon by the research organizations to provide a coherent narrative around their intellectual labor (Bowen, Rowley, Healy, & Perry, 2009). Searching within roughly a 10-year period, I collected any document or online artifact that provided insight into its self-positioning within the field. The document sources included: self-publishing and media presence, including policy briefs, reports, books, journals, website and blogs, events and public engagement, social media and media presence, as well as institutional materials, including annual reports, website, financial statements, and submissions to charity commissions or evaluation frameworks. MAXQDA was used to store and organize the information.

Analysis
The method of analysis was based on the concept of directed content analysis, also drawing on elements of discourse analysis. Theories of brokerage were used to focus the research question, and to determine the initial coding scheme and relationships between codes (Mayring, 2000). Using existing theory and prior research, the analysis began by identifying key concepts or variables as initial coding categories (e.g., "process," "structure," "outcomes"; Potter & Levine-Donnerstein, 1999). The next stage involved assigning operational definitions for each category. Here, the analysis focused on the strategies in everyday talk and text that "naturalize" relations of control (Fairclough, 1985). The application of this analysis to institutional settings and intellectual life required an examination of the structures and strategies involved in the process of self-positioning to influence intended audiences. Following this tradition, the interview transcripts and documents were coded and analyzed using MAXQDA software.
The process began with a period of familiarization with the material, achieved through the categorization/exploration of transcripts and documents, and recording variability and consistency in the data. The properties of positions taken by institutions and individuals were systematically examined, and explicit evidence was gathered by assigning segments to unique codes. The textual and contextual properties of positions taken by actors were systematically examined, and evidence for each account was coded. The codes were organized by: context/field (e.g., "university," "think tank," "media"), topic (e.g., "funding," "evaluation," "impact"), audience (e.g., "policy," "practice"), and theme (e.g., "identity," "value," "boundaries"). The second phase of analysis was concerned with identification of the functions of patterns in the process of individual or organizational intellectual labor. The extracts presented in this article were chosen as illustrative examples of the identified patterns.
This study therefore utilized an integrated theoretical framework. Interview data allowed for examination of both individual and institutional positioning, and document data allowed for examination of the way that institutions position themselves in terms of their outputs.

Setting Distances
This article argues that four distances come into play when considering the differences between contexts relevant to the production of policy research, as shown in Table 1. The ease with which each context can produce the type of knowledge that is valued and legitimate in policy and practice depends on its position in the space between fields. Yet, from my analysis, it is clear that academia occupies a particular location as an anchoring pole for research organizations of all types. That is, academia is the reference used by policy research actors to understand research, knowledge, and ideas, and as such, it becomes a key benchmark. For example, I will show how a "disciplinary distance" exists between the anchoring pole of academia and other poles (e.g., the media), whereby strongly disciplinary organizations (e.g., university departments) have further to travel from the academic benchmark to become "useful" to the world of politics and practice (which may involve a mix of political, media, and business savvy).
By virtue of its reputation and structural features, each research organization occupies a position in relation to an anchoring pole that determines how far it must travel toward salient opposing poles in the space between fields to gain credibility as useful and relevant. The following section draws on interview data; however, as will be shown in the "mapping distances" section below, this positioning is not just discursive or symbolic. It is clear that the actual production of knowledge must also shift. That is, outputs must actually be simpler, quicker, and "undisciplinary" for policy knowledge to enter policy and practice. In this way, boundaries between disciplines, professions, or contexts are not dissolved or removed, indeed their maintenance is crucial as part of the process of differentiation and association that makes up the space between fields. Rather, boundary-work establishes the distances that must be traversed. These distances, detailed below, are used to position actors in the space between fields as within the "world of ideas" or as making an intervention in the "world of policy and practice." Disciplinary Distance. A key boundary established by research actors relates to disciplines. The extent of adherence to formal disciplines works to establish distances within the space between fields. The notion of disciplines is called upon in different ways to position actors as close to either the academic pole, on the one hand, or policy and practice poles, such as politics, business, or media, on the other. For example, universities can differentiate themselves from think tanks by self-positioning along a spectrum from "disciplinary" to "undisciplinary." Being closer to the "disciplinary" side opens up academic capital, whereas being "undisciplinary" opens up capitals from other fields that confer credibility, such as, media, politics, or business.
Disciplinary organization was especially salient within university departments, where enduring hierarchies and disputes are formalized by material resources and structures, but was also observable in think tank and government contexts where contests around disciplines occur as researchers seek credibility through academic language. Particularly within university departments, one major way this plays out is a tension between economics and other social sciences. This divide was established by attention to publication outlets, the physical location of departments/centers, and prior training and material resources. For example, participants described the formal structures ("disciplined journals"), methodological differences ("quant and qual"), and philosophical issues ("positivist versus constructivist") that accompany disciplinary training. In general, academics were more likely to locate themselves within a single discipline, to attend to disciplinary demarcations in descriptions of their intellectual labor, and to lament the growing requirement to collaborate across disciplinary boundaries. Thus, within the academic context, there was a tension whereby disciplines were seen as both valuable and detrimental to policy knowledge. Here, the argument was that, if necessary, disciplines should be combined or augmented, but not dissolved, in order to produce relevant knowledge. In such cases, academics typically described research processes where separate disciplines came together at a late stage to collaborate, rather than a more integrated process from the outset. Thus, despite describing increasing imperatives for collaboration within universities, academics still undertake bounded research projects according to their disciplinary training. This shows how boundaries are set according to the logic and norms of the academic site of knowledge production. Researchers from think tank and government contexts, by contrast, were more likely to attend to disciplinary demarcations in order to describe their intellectual practice as genuinely interdisciplinary or "beyond" disciplinary concerns. These researchers lamented the preoccupation with disciplines, and advocated "getting on with it." This took two forms. On the one hand, researchers reported that disciplinary alliances continue to create tensions, but shifted the focus to finding a resolution. These researchers described the salience of disciplinary tensions, but a desire to overcome them through interdisciplinary approaches. On the other hand, a number of think tank and government researchers reported that disciplinary divides have been declining, by virtue of integration throughout the research process, rather than disciplines collaborating at a later stage. These researchers therefore largely viewed disciplinary tensions as having already been made less relevant through interdisciplinary approaches. In both cases, this focus on moving beyond disciplines is an act that simultaneously reinforces and degrades the contest over disciplinary demarcations. However, by setting up the disciplines as something "real" to be negotiated in practice, this commitment to interdisciplinarity or disciplinary integration also signals proximity to established fields in order to gain specific capitals. In doing so, researchers signal an orientation to the academic field, and thus gain credibility through association.
A related notion, often invoked by think tanks and government agencies, is "postdisciplinary" research, which moves beyond formal disciplines entirely (unlike integration). Within these contexts, disciplinary categorizations are seen as irrelevant to producing policy-relevant research and analysis. Think tank researchers often juxtaposed the value of disciplines with the value of application or impact. For example, one think tank researcher explained that disciplinary considerations take a back seat because "a lot of the work we do is quite practical, practically oriented as in managing big projects." The researcher goes on to explain that their research is "quite pragmatic" and "when we talk about research methods ... we don't necessarily look at the implications of the approach of how we see the world or vice versa." Here, being pragmatic is more important than maintaining the conceptual and methodological standards of particular communities. Both interdisciplinary and post-disciplinary discourses seek to bridge boundaries between disciplines. These strategies have two simultaneous effects. First, they position academic disciplines as irrelevant to policy knowledge. Second, they re-anchor policy knowledge within the academic field, albeit in a "transcendent" way. These allow individuals to self-position within space between fields; as closer to pragmatic, useful fields, and also, closer to the rigorous academic field.
The notion of discipline is also invoked to account for a messy, ill-defined, or "undisciplinary" field. An uneasy fit with formal or traditional disciplinary features was often explicitly attended to. Researchers described the advantages and disadvantages of a lack of true disciplinary classification. Advantages included being able to dispense with within-discipline debates to focus on practical solutions, and disadvantages included not being seen as offering serious and nuanced academic work. For example, one university-affiliated think tank researcher described: "It's certainly not a textbook sort of discipline. That has both advantages and disadvantages. It can mean there is less rigour to it. I would say that the advantage though is that you can prevent that ivory tower trap that some of the disciplines can suffer from. There is a real policy relevance to the discipline which I think is really good." This excerpt demonstrates how policy research intersects the worlds of academia, policy, and practice, but also again demonstrates a juxtaposition of discipline and "relevance." This pattern was also reflected, more subtly, through the positioning of intellectual practice to the international development field itself. It was common for researchers across contexts to elaborate the "accidental" or "fraudulent" nature of their affiliation to the field. For example, researchers would state that they were "actually" located within another discipline (e.g., economics, agriculture, or gender). The suggestion was that there was a defined field that they did not "technically" belong to.
This section has demonstrated how the notion of disciplines manifests in three ways: as contests between established disciplines, as divergent standpoints to be integrated, and as an irrelevant concern to be transcended for applied outcomes. It is thus clear that researchers across all contexts are constrained to some extent by disciplinary organization. Disciplines must either be taken up or explicitly denied. Being "disciplinary" is a key feature of serious scholarly work, and thus, crucial in positioning closer to the academic pole. By contrast, being "undisciplinary" or transcending disciplines signals useful work, for example, closer to the political pole. This section demonstrated the effects of disciplines. Self-positioning in relation to academic disciplines signals proximity and distance from different poles in the space between fields, which allows researchers to borrow capital from more established fields to establish credibility.

Epistemic Distance.
A key feature of the interview data was the concepts of research, policy, and practice, which were used to position individuals and institutions as close to either the academic pole or "world of policy and practice" poles (e.g., politics, business, or media). Participants described a distance between academic knowledge and the decision makers that receive it, related to the content and organization of evidence and ideas. On the one hand, this took the form of the complex analysis and methodologies of the scientist, and on the other, it took the form of compelling narratives preferred by decision makers. In this way, complexity of intellectual work establishes distance between the different poles in the space between fields, which can then be traversed or partially traversed. For example, think tanks can differentiate themselves from universities through positioning themselves anywhere along the axis of "simple" to "complex," whereby the former gives access to media, politics, or business capitals and the latter gives access to academic capital.
One key difference thus centers around the extent to which knowledge is "complex" or "intellectual" versus "simple" or "digestible." One academic framed the difference as arising from the value of the research, which is underpinned by opposing concepts, whereby "the benefits of research [are in] its intrinsic value and its instrumental value. You know some questions are just intrinsically important, and some are important from, you know, a policy perspective." This divide was echoed by a university-affiliated think tank researcher who drew a boundary between the two distinct modes of work required for research and policy: "For one type, you are just sitting thinking, you know, it's really policy analysis and you are drawing on your own experience … it's not a research-based process in the usual sense." The academic went on to describe how these two modes are underpinned by formal structures and incentives that influence the way the two communities are understood (i.e., "academic targets" that don't "recognise the work [done] formally in the policy field").
An example of the epistemic distance between the two worlds is illustrated by a university-affiliated think tank director: Many of the very valid criticisms of research done as consultancy I think apply to the work we have done. … [But] we wrote a book that was published by Routledge, … it had to go through the formal peer review process that any other book would do. So, we were writing something that had to be academic but also suitable for practitioners.
Here, the research is presented as rigorous through association with scholarly practices (i.e., peer review), but also as relevant or useful via being "suitable for practitioners." However, participants also described trade-offs. For example, university-affiliated think tank researchers described a need to find a balance. One think tank researcher stated: "I guess there is a trade-off. You could be spending that time doing more research, but if you are doing policy-oriented research you do need to reach out to the policy community [so you] are doing something at least relevant." Here, the trade-off centers on time invested in projects, which belies a judgment around relevant policy work over disengaged social science.
A related concept corresponds directly to the role of the knowledge broker. In this conception, the academic inhabits the ivory tower and is concerned with the self-referential community that produces and circulates knowledge according to its own standards. The think tank researcher, by contrast, is a go-between that translates scholarly production into policy-relevant knowledge. This academia-policy divide was illustrated by a think tank researcher: "It is actually really our role to be that go-between. … Some people are getting to the research fellow level although they absolutely don't know how to do research, but they are really good at giving ideas and transmitting ideas. … We have to find a new way to exist, I think, between all that." Similarly, a government researcher describes intersecting communities involved in a flagship publication, which is designed for "the development community, so more practitioner, policymaker oriented than research, but it's often run within the research department." The respective communities thus come to be understood in terms of the nature and function of ideas. In academic contexts, once ideas are properly vetted and critiqued by the academic community, they can be translated and used by the policy or practice community. In think tank and government research contexts, by contrast, ideas obtain value in their ability to meet the needs of policymakers and practitioners in order to solve social, political, or economic problems. This is illustrated by a think tank researcher who described the problem of translation: Most of the academics are doing quite fine-grained research at basically project level in a particular location in a particular country. … That's all fine, but even at that level, there doesn't seem to be any mechanism for feeding that back into the design of aid programs. … There is no real twoway engagement between the academic sector and the sort of public sector which extends out to include the different types in most countries.
To translate their findings, policy researchers across contexts often moved away from the expert persona and positioned themselves as brokers of knowledge and ideas. In this way, researchers can appropriate characteristics from available sources and employ them as necessary. A government research director (and former academic) illustrated the interaction between the goals and modes of production of each context: "When I was in academia as a consultant and as an advisor [to the government and multilateral donors], I felt that one of the things that I can really contribute on, all the things where I can be engaged and informed, is the real world of policymaking." This exemplifies numerous accounts of individuals crossing professions in my sample. Researchers use these categories and identities to position their intellectual labor and their ideas. Thus, there is crossover between professions, both in terms of prior employment, but also, in the construction of relevant actors. Given the highly networked nature of the field, actors position themselves in relation to other contexts, both establishing and traversing the distance between the goals, concepts, and practices of each.
There is therefore a conceptual divide between the epistemic worlds of "policy and practice" and "research." The deep knowledge held by academics is positioned as separate from the practical orientation of recommendations for policy and practice. As such, there is ongoing negotiation of what is possible and credible in terms of the realities faced by the policy community (i.e., electoral, political, financial, and pragmatic constraints). Researchers seek a competitive point of difference by asserting the scientific quality of their work, while attending to "real" challenges in policy or its implementation on the ground.
Temporal Distance. Another boundary often drawn by interviewees relates to temporal differences in the production of research. This worked in opposing directions. On the one hand, researchers placed value on the quality and rigor of slow, meticulous work. For example, some think tank and government researchers bemoaned the lack of time to conduct "proper research," locating their production in scholarly terms but with a clear temporal difference. On the other hand, researchers placed value on the timeliness or responsiveness of their work. Think tank and government researchers positioned academic research as too slow to be useful to decision makers. Thus, this distance centers around the extent to which knowledge is "slow" or "rigorous" versus "fast" or "agile." In terms of respective positioning, the academic is involved in intellectual production that is slow because of its epistemic qualities (e.g., highly critical, technical methods), and the policy researcher is involved in intellectual labor that must anticipate or respond to current events in time to make a true intervention.
This temporal distance is illustrated by a university-affiliated think tank researcher who described the difference between the production of "types" of outputs: "there is an immediate need to get the research out to policymakers through, [and] that's important but the questions that policymakers want to know about might be different than the kind of rigour that would apply to getting something publishing in an academic journal." Another think tank researcher described, based on previous experience that international agencies "often do things in a hurry for short-term contracts and they may or may not have the expertise in the area that they're rushing on that week, but there's an imperative to produce and get outputs out the door." This echoes the view of a think tank researcher, who described the difference in timelines between academic projects and commissioned projects: "In terms of thorough research, we would like at least three years of research [but] you actually have nine months. ... You can't ask research to solve all those things if you don't allow research to have the time." This is reiterated by another consultancy-type think tank researcher, who describes a "split" organizational profile, which juxtaposes slow academic work and fast policy work: "I would say that different parts of [the institution] behave in different ways at different times, so there are parts that basically behave like an academic institute, you know, they will do longer work, and there are parts, which would take longer to do good research." Other researchers used temporal notions to position academics as overly pedantic, leading to irrelevance. Slow, careful attention to method, theory, or subject is thus sometimes desirable and valuable, but at other times impossible or undesirable. This divides "agile" policy outputs from slower academic work. As a multilateral researcher described: "you are working with limited resources, with limited policy options, and limited time. … Sometimes the constraints come from pressure on the policymakers, that everything has to be done instantly to get instant results." Here, instant results are required for influence. Similarly, a university-affiliated think tank researcher describes the time constraints that arise from being involved in commissioned projects: "Having finished a major contract [and] written it up, then you have a very short period of time it has to be reworked and revised to put into publishable form and that's something we don't often achieve. We just move on from one contract to the next." These accounts, while undoubtedly reflecting genuine time constraints, also position intellectual labor as oriented to the needs of decision makers. Thus, a boundary is established where fast, agile work is "relevant" and slow, academic work is "irrelevant." What is notable, however, is that this conception of university researchers as having "adequate time" to undertake "proper," "meticulous" research is not borne out in academics' accounts. Rather, academics describe time pressures, including "sourcing research funding," "mobilizing research consortia," and "rushing to publish." Academic timeframes may be perceived as more generous, but time pressure is a feature of contemporary research across contexts. Time constraints arise from varying factors, from meeting the requirements of funders (and evaluations) to anticipating policy needs. Temporal notions thus achieve different effects, locating research as either "rigorous" or "relevant." Statements of this type can be understood as insight into structural conditions of a particular context (e.g., pressures of contracts and funding arrangements), but also as a demonstration of relevance and usefulness.
Economic Distance. A further boundary often drawn by interviewees relates to differences in modes of employment. There are clear boundaries that exist between "pure" research supported by core funding, and "marketable" research supported by various types of fundraising. However, the wide variety of funding modes makes this a complicated exercise in positioning for individuals and their institutions. Many researchers from all contexts are engaged in consulting projects, adopting a professional role as expert advisor on short-term contracts. Consultancies and contract research is thus a major way scholars can make, or position themselves as making, political and practical contributions. However, there is a trade-off in this potential for influence, whereby there is a perception that contracts quash independent thought.
It is notable that although many researchers reported taking on consulting projects or working in a commissioned research style think tank, none positioned themselves as consultants. Rather, consulting arrangements were seen as a mode of employment (albeit with specific consequences for intellectual practice), but did not constitute an "identity." In all cases, the positioning centered around credentials in the academic, policy, or practice fields, rather than the type of employment. For example, an academic set a boundary between current scholarly work and previous work as a consultant: "I am in academia now, so something that occupies more of my thinking than when I was in practitioner circles is the misuse of research among practitioners." In this way, the researcher sets their identity within the academic field, and creates a boundary between the two worlds of research and practice.
This boundary-work is also evident in the following extract from an academic who describes the relationship between research and its users in developing countries: For me as a researcher that's really critical, that relationship, but at the same time, having that relationship then influences my research and means that I become more limited by the practice environment, and not thinking as broadly as I perhaps should, because I'll be influenced by the practitioners of the area. ... So, having to keep that in mind, but still, you know, there's limited time and I want to be a useful academic.
Here, consultants provide "useful" knowledge, drawing on scholarly credentials to provide a pragmatic understanding of policy and practice. Consulting thus can make research valuable, relevant, or impactful. Interviewees across contexts described seeking out alternative sources of funding as public funding diminishes, which corresponds to the commercialization of research. Private sector language is used to describe intellectual labor, for example, "find the market for the knowledge," "outsourcing," and "products." This consulting and evaluation research is core business in think tank and government contexts, and more marginal in universities. Many researchers consider consulting a key part of their work, others described it as a side activity. Across contexts, interviewees described the worlds of research and consulting as two points on a spectrum, where the former is slow, rigorous, and complex and the latter is fast, accessible, and practical. In all cases, commissioned work is seen to be shaped more directly by funder priorities than academic research, requiring researchers to employ different strategies to protect their cognitive independence. The narratives drawn on by researchers also demonstrate the ways professions are enacted. Interviewees within each context also attended to the practices, identities, and positions of other research contexts. For example, an academic described the benefits of simpler, mixed-methods consulting work, but repositioned himself as firmly academic: People in other areas in development, other disciplines have heavily criticised this empirical method [on] ontological and epistemological grounds and fair enough. That's another reason why I like the mixed-method approach we've been using in the consultancy work, although I am fundamentally an academic [and] my reputation depends on my ability to publish in journals.
This distance was also demonstrated by a university-affiliated think tank researcher, who used academic qualifications as an explanatory factor for intellectual freedom (i.e., setting their own agenda), and thus, by implication, to retain cognitive autonomy against the interests of funders. So, with a PhD you have a greater degree of freedom [and the] positions that you take are mostly up to you as long as you can defend them. And in the policy realm [it can be] very difficult to get people to understand that as a consultant you're not representing the organisation, you're bringing expertise to write a report about findings and research that both parties are interested in.
One effect of this account is the positioning of intellectual work as neither entirely academic nor entirely economic. There are strategies available, such as highlighting philanthropic funding, that allow genuine policy influence (i.e., unlike academics), without being beholden to funders (i.e., unlike consultants).
The distance was also displayed in government research. Oriented to politics and policy, government researchers drew on academic, consulting, and international professions to locate their intellectual practice. For example, one government research director demonstrated the complexity of positions required to access various capitals: "For a consultant, you want feedback, your performance metric is sustainable profit. For [government agencies], it's the empirical quality of your product and implicitness of the stakeholder consultation." A key demarcation between "practice," "policy," and "academia," therefore relates to a distance between "pure" intrinsic research and "marketable" products. As such, economic value or marketability has become a key indicator of "usefulness." This economic distance positions rigorous scholarly work as a way of resisting the influence of the employer, donor, or funder, but simultaneously, legitimizes consulting or third-party funded work as useful, impactful, and thus, valuable to the "world of policy and practice." Thus, particular disciplines, professions, and contexts are used to situate the intellectual production of actors through boundary-work.
By setting out four distances, this article has demonstrated how the space between fields is constructed through boundary-work. The next section considers how this boundary-work structures knowledge production, by mapping the distances within this space.

Mapping Distances
I have argued that there are four key distances in accounts of knowledge production in policy-oriented contexts. These distances are constructed through an ongoing process of differentiation and association, which locate an actor in the space between fields. Individuals and institutions are free to position themselves in various ways from moment-to-moment, but a meta-strategy must also be adopted to find a balance between distances. Structural and historical constraints, contained within an existing identity, make it easier or more difficult to self-position in certain ways. The notion of distance thus becomes useful in understanding the relative "difficulty" or "ease" of this positioning. That is, instances where an actor has to work harder to present themselves as belonging to a particular field involve greater distance, and instances requiring less effort involve shorter distance.
In order to show this concept of distances graphically, I have adapted Medvetz's (2012b, p. 37) "Think tanks in social space" diagram to illustrate the development research space. Figure 1 places the institution types within the space between fields.
Overlaying the case organizations investigated in this study, Figure 2 represents a rough "positioning map" based on quantitative publication analysis and qualitative analysis of interviews and institutional documents. The location of the organizations on the map thus captures something of their respective identities and their structural features such as prescribed routines, practices, or constraints.
The results of my study suggest that for each distance the anchoring pole or "distance marker" is the academic field of knowledge production. This reflects the site of "ideas," which has traditionally involved "disciplinary," "complex," "rigorous," and "pure" modes of intellectual labor. By contrast, in this example, the economic, political, and media fields correspond to the "world of policy and practice," which each require (various combinations of) "undisciplinary," "complex," "direct," and "marketable" knowledge. For example, to gain political capital, one might emphasize that their intellectual work is politically "useful" because it is fast, straightforward, and not mired in disciplinary debates. Any policy research organization's first goal, even prior to that of exercising influence on policy or practice, is to differentiate itself from its parent institutions (Medvetz, 2012b), thus there is always a closeness to the academic sector but always a concomitant distance.
There is no assumption of evenness between these distances. Strategies to move from "ideas" to "action" can operate independently or in combination, and toward one or several parent fields simultaneously. For example, a think tank researcher might emphasize their "real world" relevance by describing the organization's highly responsive blog (addressing only the temporal distance). By contrast, over the course of a conversation, a researcher may invoke all four distances (and potentially others) to achieve multiple aims from moment-to-moment. The key point here is that identities and positions are not fixed, and as such the distances (i.e., the amount of positioning work required) shift from instance to instance.
Although identities are not fixed, there is a fundamental coherence between structure and identity in the space between fields. Researchers comprehend their social role in terms that correspond to their location in social space. As Medvetz (2010, p. 550) notes, "lacking an established definition of what it means to be a policy expert, such actors typically improvise one using the ready-made cultural materials supplied by the more established institutional domains to which they are linked." I provide visualizations of conceptual distances and concomitant strategies within each context below. Figure 3 illustrates how the four identified distances might work for a more "traditional" university department, using evidence from the Oxford Department of International Department (ODID). To gain academic capital, the department and its researchers improvise an identity primarily from the most salient institutional domain/parent field (i.e., the academic field), conceptualized here as a short distance, because the "disciplinary," "complex," "rigorous," and/or "pure" nature of their intellectual labor is taken as self-evident. However, when seeking to establish credibility in the world of development policy and practice, this same self-evident quality involves a greater distance to political and practical utility. This necessitates positioning strategies that emphasize the "undisciplinary," "direct," "agile," and/ or "marketable" aspects. For example, an academic might emphasize their involvement with an interdisciplinary, easily understood report for an external agency. Examining the websites and institutional materials of organizations in the space allows us to see this in action. For instance, ODID (2015, p.16) states "while emphasising academic rigour, our research engages explicitly with policy issues -albeit critically and with a long-term perspective." Here, there is a short temporal and epistemic distance by virtue of its structural proximity to the academic field (i.e., existing as part of the university), and greater temporal and epistemic distances to "contribute to better design and implementation of development policy and practice by both government and non-governmental organisations" (ODID, 2015, p. 19). Figure 4 illustrates how the four distances might work for a university-affiliated think tank, drawing on examples from the case of the Institute of Development Studies (IDS). IDS has a reputation for high-quality research, teaching, and consulting work. As such, it is relatively easy for the organization or its researchers to gain academic credibility (via pointing to academic-style outputs, research fellows, physical proximity to the university, etc.). This is thus a mid-range distance to the academic pole, which can be traversed by emphasizing the "disciplinary," "complex," "rigorous," and/or "pure" nature of their intellectual labor. However, this association with university production potentially makes it difficult to gain political capital. This could also be conceptualized as a mid-range distance, which can be traversed by pointing to the "undisciplinary," "direct," "agile," and/or "marketable" engagement with stakeholders and consulting for key agencies and so on. A statement by Lawrence Haddad, former IDS director, illustrates an "in-between" institutional position: "IDS occupies a unique space between think tank and university [which] reflects widespread perceptions that we are one of the world's leading policy engaged academic institutions" (IDS, 2013). Figure 5 illustrates the four conceptual distances in relation to a government site, utilizing evidence from the case study of the World Bank. The World Bank Research Group is an interesting example because it is a dominant political actor, which also seeks to position itself as a media-savvy, publicly accountable, "knowledge bank." Given its strong political reputation, self-positioning as politically useful requires little positioning work or a "short distance" (i.e., the Bank already has the structures and accumulated capital to make political interventions). On the other hand, self-positioning to access academic capital requires more effort, and thus involves a greater distance (e.g., meticulously detailing "disciplinary," "complex" methodologies, "pure" apolitical topics, and "rigorous" extended research projects). As the "Research at Work" statement states: "Bank researchers produce a large volume of work that is of high-quality and influential by academic standards, yet much more focused on development issues and developing countries when compared to the research of academic institutions" (DECRG, 2015, p. 3).
This section considered how the disciplines, professions, and contexts are constructed through boundary-work, which demarcates the limits of acceptable research. This boundary-work sets four distances that signal proximity to different available poles, such as academia, business, media, and politics. This is an uneven process, and the association of research with traditional academic sites exerts a strong structuring power. As such, the academic field is the "distance marker," allowing positionings that are either close to the "world of ideas" or closer to (one or more) "world of policy and practice" poles such as politics, media, and/or business. These positionings can vary within individuals, organizations, and within contexts, and differ in terms of their "coherence" (Williams, 2018). Each context thus provides broadly distinct products and ways of knowing. However, the findings also suggest that positions overlap; each actor has a particular way of seeking, compiling, and shifting knowledge in relation to disciplinary, epistemic, temporal, and economic distances.

Conclusion
I have argued that policy expertise is constructed through boundary-work that locates it in the space between multiple disciplines, professions, and institutional contexts. I developed the concept of "distances," and explored the ongoing negotiations that allow actors to navigate them. Through ongoing positioning, actors, ideas, and techniques can travel between multiple fields (Medvetz, 2012b). This liminal quality confers certain benefits (Eyal, 2006;Medvetz, 2012a). As Stampnitzky (2011, p. 3) argues, weak boundaries permit those on the boundaries to draw on and appropriate diverse ideas and approaches, and to seek out the routes to esteem and influence provided by parent fields. This in turn facilitates the reach and influence of those ideas, and permits faster dissemination than if they were contained within established fields. Examining the ongoing positioning of research contexts to traverse distances shows how knowledge brokerage relies on various kinds of boundary-work that structure knowledge production.
Although "pure" professionals from more established fields (e.g., politics, academia) are found within each context, in policy-relevant settings like international development, permeable borders mandate the development of hybrid intellectual skills and practices across all contexts. In this liminal field, researchers from all contexts must self-position as knowledge brokers, despite "natural locations" within universities, think tanks, or government. The traditional notion of field provided by Bourdieu (1993) implies a relatively bounded space, with a concomitant set of symbolic struggles. However, the approach employed here showed that particular institutions do not occupy fixed separate locations within the space of policy knowledge by virtue of their overarching context. Rather, universities, think tanks, and government research organizations are a part of a diverse boundary location that contains multiple participants involved in policymaking and practice. Actors from each context must mediate between a host of ideas, aims, and approaches, given no one discipline, field, or profession has a monopoly over authority or credibility.
The techniques and strategies of knowledge brokers appear across research contexts, from those closest to the academic pole to those in the center to those at the political pole. However, by virtue of boundaries given by structural elements (e.g., funding and evaluation), each context has different criteria for valued and credible research. Indeed, the processes and practices for idea generation and dissemination do vary within and between contexts. The article offers an exploration of tensions that delimit acceptable types of knowledge. In doing so, it unpacks the concept of the knowledge broker as situated between two worlds' policy and practice, and shows how these worlds are instead invoked by research actors to achieve various aims. This work provides a new vocabulary for describing relational proximities through the language of distances. Thus, unpacking these worlds through empirical examination of the distances between types of knowledge offers a new way of conceiving the interaction of disciplines, professions, and sites of production in the service of credible expertise. Therefore, the performance of knowledge production, and whether research actors fall more toward particular fields (e.g., academic, political, media, or business) reflects strategies of language and action. These strategies structure the shared space between various research contexts. In this way, the acceptance and dissemination of intellectual products depend not only the strength of the argument or evidence, but also on the set of devices employed by individuals to position themselves within established fields, and indeed, the spaces between them.
The argument that policy experts maintain their credibility by playing off different disciplines, professions, and sites of knowledge production in their pursuit of a range of capitals has a number of potential implications. It provides a new way for policy researchers to consider how knowledge and ideas are used and translated. Researchers gain symbolic power through their ability to offer "useful" knowledge (i.e., "undisciplinary," "direct," "agile," "marketable" outputs), thus actors may wish to map their capital profiles by tracing their intellectual interventions (i.e., to demonstrate engagement and participation in policy). This may provide insights into sources of untapped potential capital (e.g., a bureaucratic government agency could pursue a media-oriented strategy) and offer a deeper, potentially alternative, understanding of "evidence" for the influence of an organizations' interventions. For example, research contexts toward the academic pole may seek to make more use of their power and status by making short pieces of research freely and publicly available via a central publications website or blog. Those at the center may wish to implement and communicate formal peer review processes or networks for outputs of various types in order to benefit from the reputational capital afforded to rigorous impartial evaluation. Those toward the political pole may wish to reframe "evaluation" in the language of academic research or formalize practices that perform individual ideological autonomy. Research contexts could also each bring their respective "natural" skills (e.g., academics' deep knowledge or think tanks' mediation skills) to actively engage with, critique, and offer alternatives to the content produced by those at other locations in the space between fields. This represents a refocusing on the content of diverse types of knowledge, rather than on respective prescribed roles within the space, which provides opportunities for researchers of all types to restate and assert their positions within the knowledge production system.

Kate Williams is an ESRC Future Research Leaders Fellow in the Department of
Sociology at the University of Cambridge and a research fellow at Lucy Cavendish College Cambridge. She is also a postdoctoral fellow at the Weatherhead Centre, Harvard University. Her research focuses on the production, use, and evaluation of policy knowledge. Note 1. The salient arenas of international development that participants drew upon were academia, policy, and practice. Although the lines between policy and practice were sometimes set out explicitly (e.g., operational staff "on the ground" constitute "practice" and professionals within agencies constitute "policy"), for the most part participants conflated the two. They talked of policy and practice as both being concerned with impacting decision makers in the "real world." This is a proxy provided by the funding breakdown for the university as a whole, given the lack of designated department and corresponding lack of funding information.