Introduction

The ‘gap’ between science and policy, seen as the inevitable consequence of the fact that both represent different cultures or even worlds, has been extensively studied, debated and criticised in the field of science and policy studies (Wehrens et al, 2011; van der Arend, 2014; Wellstead et al, 2018; Mols et al, 2020). Historians have deconstructed how the spatial metaphor of a ‘gap’ highlights and/or constitutes alienation, separation and distance between (the practices and networks of) scientific research and ‘others’, like policy, but also industry and the public (Bensaude-Vincent, 2001; van Rooij, 2013). More in general there is a wide-shared understanding that the use of evidence in policymaking, and the relation between scientific research and political decision making, is not self-evident, potentially problematic and has its own politics. Many studies of the research-policy interface highlight primarily the complexity of the policy process, emphasise the role of individuals, and reduce the research process to a ‘use/non-use’ binary (Parkhurst, 2016; Neal et al, 2022). Consequently, concepts like ‘coordination’, ‘boundary spanning’ or ‘cross-domain orchestration’ are invoked to study how ways of working, temporalities, aims and interests are aligned to ‘bridge the gap’ (Miller, 2001; Bijker et al, 2009). But such concepts often remain ‘nebulous’ (Bednarek et al, 2018) or all-together ill-defined (Neal et al, 2022), and do not contribute to an overarching perspective on cross-boundary knowledge work (Järvi et al, 2018). Thus, calls in the literature still appear for analytical alternatives to a simplistic and persistent ordering in terms of a linear model (Maas et al, 2022), as well as more attention for the legitimacy of the process and the role of (knowledge-brokering) organisations (Parkhurst, 2016; Wye et al, 2019). In response to this ‘gap’ in the literature, we present a multidimensional and non-linear analytical framework that can be used to describe and interpret various types of coordination work that take care of the messy realities in between science and policy.

In this paper, we develop a proximity approach to the coordination work and productive interactions between science and policy. We take as our empirical focus structural, institutionalised interfaces between science and policy, what have been called boundary organisations or ‘hybrid research spaces’ (Wehrens et al, 2014). To better grasp coordination efforts in these spaces, we turn to studies in the geography of innovation that are motivated by a similar concern for coordination, but then in knowledge-exchange mechanisms between companies and knowledge institutions (Boschma, 2005; Alpaydın and Fitjar, 2020). Hence, we introduce the multidimensional proximity concept to make visible which actors, interests and needs reciprocally structure the functioning of science-policy collaborations (Torre and Gallaud, 2022a). Ultimately, this makes room for understanding (and comparing) the politics of producing and using evidence for policymaking (Parkhurst, 2016). Similar to recent work on knowledge co-production in ‘protective spaces’, this can contribute to the development of effective organisational forms for research-policy interactions (Boon et al, 2019).

Therefore, the main research question is: how does proximity in hybrid research spaces enable coordination between science, policy and practice? We take ‘productive interactions’ (Spaapen and Van Drooge, 2011) as qualitative indicators of successful collaborative knowledge work. We develop the proximity framework through a site-specific case from the city of Rotterdam, in the Netherlands, where we studied a ‘knowledge infrastructure’ consisting of city government sections and local knowledge institutes working together in various hybrid research spaces. The paper is structured as follows. First, we provide a theoretical background by discussing the literature on coordination work, proximity and productive interactions in the context of the collaboration between science and policy. In the Methods section we describe our approach and the set-up of our qualitative research for the Rotterdam case. In the Findings section, we present the results of the proximity approach for this specific case. Finally, we reflect upon the politics of coordination work in science-policy collaborations, as laid bare through the proximity approach. We conclude with limitations of the study, lessons for collaborations elsewhere, and suggestions for further research.

Background

Here we outline the theoretical background of this study. We briefly discuss the current literature on, respectively, coordination work between science and policy, proximity, and productive interactions.

Methods

The current paper results from an evaluative research project focused on a system of local science-policy collaborations in Rotterdam, or what participants called a ‘knowledge infrastructure’. This infrastructure can be understood as a (connected) series of hybrid research spaces. We focused on eleven collaborations that were explicitly recognised as part of the formal collaboration between university and city, both on the university’s website as well as by key actors. The evaluation project was conducted at the request of the primary sponsors of the collaborations, the Rotterdam municipality and the Erasmus University Rotterdam. Here we present a sub-study that was conducted within this project, focusing on proximity dimensions as conditions for productive interactions. We therefore only elaborate here on the methods we used within the sub-study. An in-depth description of the evaluation project can be found in the publicly available report (Smit et al, 2021).

We used a combination of qualitative research methods for our data collection. We conducted 45 semi-structured interviews varying in length from 45 to 60 minutes (some of which were duo interviews). We made a respondent selection from 11 science-policy collaborations (see Appendix for an overview of the case selection), ensuring a balanced representation in terms of professional background (municipality, university and other organisations) and role within the collaboration (for example, coordinators and researchers). We used an interview protocol that focused on the productive interactions between heterogeneous actors, and specifically invited respondents to pronounce evaluative statements to make visible potentially conflicting views on what constituted successful cooperation. The protocol was thematically structured, covering the organisational structure, composition, interactions, outcomes, strategy, visibility, and vision for the future.

Two focus groups were organised to corroborate and enrich findings from the interviews. The strength of focus groups lies in the group dynamics that stimulate participants to clarify their views and motivations (Barbour, 2008). We used a semi-structured format, inviting participants to elaborate on, and respond to, examples of productive interactions and impact to gain a deeper understanding of the enabling conditions. In a second part, participants collaboratively visualised a future vision of the knowledge infrastructure. This interactive method stimulated active engagement and discussion between participants and helped to make ideas about the spatial and organisational arrangements that constitute the infrastructure, as well as key actors within it, more tangible and concrete (Sanders and Stappers, 2012). The focus group participants were selected from our interview sample, apart from two who were not interviewed beforehand due to staff changes. We used a select composition to guarantee sufficient diversity (of different collaborations and institutions) and similarities in organisational structures to allow for mutual recognition and comparison. In the selection we focused on individuals with a strategic perspective (such as coordinators) that have overarching insight into the practices of the collaborations and the core principles guiding them.

The composition of our total respondent sample can be found below, in Table 1, where we paid attention to the hybrid composition of the collaborations. The table is structured in the following categories: municipality, university and third parties. University includes academic staff and researchers at EUR-affiliated contract research institutes. Municipality includes policy advisors and researchers. Third parties includes researchers from other universities (of applied sciences) and private actors.

All interviews and focus groups were conducted in the period between April and June 2021 and took place online, due to pandemic restrictions, via videoconference software (Microsoft Teams). The protocols for the interviews and focus groups are available through the Open Science framework (Smit, 2023). The audio recordings were transcribed and subsequently coded in Atlas.ti. We took a thematic approach to coding where we distinguished four analytic code categories (activities, outcomes, impact, conditions) to which we inductively added codes, which were altered and combined to limit the number of codes. In this paper we focus on the conditions for productive interactions. In the analysis stage, we associated various conditions with the four dimensions of proximity, in order to identify barriers and enablers for coordination and interactions. In doing so, we provide, in Table 2, (still much needed) qualitative operationalisations of proximity dimensions (Alpaydın and Fitjar, 2020).

Findings

In this section we discuss the way in which different proximity dimensions mattered to productive interactions in the hybrid research spaces. Illustrated with quotes from the interviews and focus groups, we demonstrate how the proximity approach allows a fine-grained understanding of the conditions that facilitate and prohibit productive interactions and knowledge exchange. Per proximity dimension we report respectively on relative significance, benefits to knowledge exchange and productive interactions, supporting factors and potential unintended effects. Mostly we focus on the experience of ‘university’ or ‘science’ and ‘municipality’ or ‘policy’ actors, only highlighting differences within these categories when this was relevant in practice. In conclusion, we summarise these findings in Table 3.

Discussion

In Table 3 we summarise our findings about coordination work between science and policy in terms of different proximity dimensions. As a whole, it agrees with many earlier findings that highlight the ambiguity or messiness of the science-policy interface. It turns out that coordinating the front- and back-stage of collaborations, as well as the different organisational dynamics between scientific research and policymaking, can be described in terms of different degrees and combinations of proximities. This also always includes distancing, because at various instances, proximity could be both beneficial and detrimental to productive interactions. Although the overlaps and substitutions between proximity dimensions might make unambiguous differentiation not always possible, a proximity approach has allowed a further articulation of the conditions for productive interactions in hybrid research spaces. That is, proximity can function as vocabulary for actors’ dealing with messiness at the science-policy interface.

Although, above, we have analysed a diversity of science-policy collaborations together, one can deduce from the overview of involved actors, in the Appendix, that there is not one straightforward mechanism to overcome coordination problems or to do ‘boundary management’ (Miller, 2001; Wehrens et al, 2014). Indeed, we learned from our interviews with diverse actors that each collaboration combines the proximity dimensions in different ways, which typically results from historical developments and functional motivations. For example, one collaboration explicitly used geographical proximity in a shared office space to bridge cognitive and organisational distance, while another collaboration relied on long-standing social proximity to enable productive interactions. This shows, in addition, that proximity dimensions are interdependent insofar as they influence and can even reinforce each other (or overlap), and that the same outcome (productive interactions) can be achieved by different balances between dimensions (or substitute) (Hansen, 2015). Practically, this implies that actors in hybrid research spaces can also use the proximity approach to strategically nurture specific dimensions, in order to improve productive interactions and achieve their goals.

This brings us to the reflection that we can interpret an analytic difference between two ways of enacting proximities in our study of science-policy collaborations in the Rotterdam hybrid research space. Geographical and organisational proximity appear as practical contexts in which the social and cognitive dimensions of proximity can be performed, reinforced or reduced. Whereas most proximity literature treats at least the non-spatial, if not all, proximity dimensions as analytical equals (Boschma, 2005; Heringa et al, 2014), our employment in this science-policy case indicates that we should actually pay attention to the asymmetrical relations between different proximities, especially in practice.

Building on this analytical distinction between more ‘ordering’ proximities (organisational, geographic) and ‘operating’ proximities (social, cognitive), we claim that there is a politics of proximities (Smit, 2021) at play in the coordination between science and policy practices. It is the case, namely, that each (more or less consciously developed) mix of proximities implies a dynamics of inclusion and exclusion with respect to the collaborative epistemic process. Organisational and geographic proximity shape to a large extent which knowledge, perspectives, interests and relations can be accommodated and to what extent. Below, we describe the difference between ordering and operating proximities therefore in terms of a material-epistemic (where, how) and a socio-epistemic (who, what) politics of coordination work.

First, socio-epistemic politics of coordination determines the who and what of a hybrid research space: which actors and organisations are included in the formal and informal networks, and what knowledge, experience and skills are recognised as legitimate and relevant. In this way, social and cognitive proximity shape to a significant extent the agenda, problematisation, methods and dissemination of the collaborative research. In the Rotterdam hybrid research spaces this is most tangible in the closeness between social and medical scientists, on the one side, and highly-educated policymakers, often with similar disciplinary backgrounds, on the other. Many of the research spaces are quite selective epistemic communities, in which the participating scientists tend to shield their network, keeping actors from other disciplines or institutions at a distance. This excludes possible alternative knowledge perspectives on a problem, and can prohibit access of other researchers to policy. In addition, actors report that they can be more critical and conflictual after high trust is achieved, which can mean that other, potentially more critical, perspectives have been excluded. Thus, the front-stage celebration of ‘fluent’ communication and ‘fast’ response to urgent issues could invisibilise back-stage exclusions and frictions (Tsing, 2004).

The socio-epistemic politics of coordination work did, in this case, not so much exist in the explicit unethical use of science to make political arguments (‘technical bias’), apart from politicians at times wilfully ignoring a report. Rather, we did encounter the ‘subtle politics’ of evidence-based policymaking (or ‘issue bias’, Parkhurst, 2016), that creates a limited perspective on what counts as political and scientific problems and possible solutions. On several occasions we observed that collaborations align their agenda strongly with the political priorities of an alderman. And, although many collaborations aim to, or do, include also third ‘practice’ parties, the inclusion of actors outside of policy and science is typically experienced as more difficult. By implication, relatively powerful policy and science priorities end up shaping the agenda, ways of working and, ultimately, allocation of resources. As mentioned above, mediators, hybrid actors who move easily in both worlds, might become powerful actors in such settings.

Second, the material-epistemic politics of coordination work denotes the where and how of a hybrid research space: its location and space, organisation and institutionalisation affect who can participate, and what knowledges can be developed. Spatial and organisational arrangements are no neutral backgrounds but structure what interactions are possible, which actors can participate, in what ways, and what makes legitimate issues, approaches and solutions (Henke and Gieryn, 2008; Gieryn, 2018; Smit, 2021). In the Rotterdam hybrid research spaces, this played both at a micro and macro level. Within the collaborations, there is a politics to deliberation: who decides how to decide, and where do the deliberations take place (who has to come to who)? On the other hand, it became evident that the creation of organisational distance to perform independence can be a way to desituate knowledge and exclude dissensus and alternative perspectives, to make it look ‘objective’ and ‘scientific’, while at the same time shaping it to the needs of formal political deliberation.

At a macro level, we observed this politics especially with respect to the position of the ‘ordinary Rotterdam citizen’. It might be surprising that we have been able to refrain from discussing inhabitants of the city, or more precisely, an ideal-type citizen that most collaborations considered their ultimate beneficiary. This is, however, an artefact of the situation that few organisational or spatial measures were in place in the hybrid research spaces to involve these actors or monitor the effects for them. The relation to the Rotterdam citizen was, thus, more of a ‘front-stage’ affair by mentioning them as ultimate beneficiary in policy documents and organising events in public spaces and inviting them. The fact that several actors could refer to the city as a ‘laboratory’ is a sign of this spatial-organisational exclusionary dynamics. This paradox was especially tangible in the focus groups, where we invited participants to imagine an ideal spatial and organisational arrangement for the knowledge infrastructure ‘in 2025’. Whereas ‘the citizen’ or ‘the city’ was typically put centre-stage in their spatial imagination – be it in concentric circles, a metro line or a neighbourhood market – they at the same time acknowledged the difficulties of achieving this in practice, instead admitting that their collaborative work remained tied to the existing policy agenda and their habitual research methods.

In conclusion, we thus propose to distinguish between two types of epistemic politics at play: the material-epistemic politics of ordering proximities and the socio-epistemic politics of operational proximities. Thereby, the proximity framework introduced in this paper does not only allow us to analyse different types of ‘solutions’ to the coordination between science and policy practices, but also helps to highlight the infrastructural embedding of power differences and exclusionary dynamics because of the ‘politics of proximity’. Notwithstanding attempts to characterise different boundary management strategies (Wehrens et al, 2014), such a comparative and analytic focus had been lacking and the proximity literature has disregarded science-policy exchanges as well as the asymmetric interplay between proximity dimensions.

Practically, this implies that to change the socio-epistemic politics of hybrid research spaces – the topics that are studied, the knowledge that is used and produced – one can focus, first, on the ordering (spatial, organisational) aspects of science-policy collaborations that stimulate operational (social and cognitive) proximity in the desired directions. From our case study, we have summarised suggestions in Table 4.

Table 4:

Case-based suggestions for the stimulation of productive interactions through proximities

View Table

Conclusions

The main question guiding this research was: how does proximity in hybrid research spaces enable coordination between science, policy and practice and, thus, cultivate productive interactions? We studied a system of local science-policy collaborations in Rotterdam, a ‘knowledge infrastructure’, which we interpreted as a collection of hybrid research spaces. Taking this approach allowed us to distinguish how different proximity dimensions mattered to coordination between science and policy, on the front- and the backstage. In doing so, the proximity approach provides the instrumentarium for a fine-grained analysis of the enabling conditions (de Jong et al, 2022) that facilitate and prohibit productive interactions and knowledge exchange. Proximity can be used as a general analytic framework for understanding coordination work and collaborations at the science-policy interface (Järvi et al, 2018; Wye et al, 2019; Maas et al, 2022). Although the different dimensions influence each other, we showed that it is valuable to make an analytical distinction between them, as operational and ordering proximities shape in different ways productive interactions and can be managed in their own, specific way.

In general, we conclude that the attempt to facilitate productive interactions by organising hybrid research spaces can be seen as a successful strategy to deal with the perceived social, epistemic and organisational distance between science and policy. The various knowledge labs established proximity in different dimensions and combinations, which often contributed to the (perceived) ease with which researchers and policymakers can find each other and collaborate.

However, we have also highlighted how social-epistemic and material-epistemic politics of proximity not only establish productive interactions between specific actors, but also – by implication – can enlarge the gap with others. This implies that the meaning of interactive ‘productivity’ is itself open to (political) debate (Godin, 2009), something that few in this literature explicitly acknowledge (Spaapen and Van Drooge, 2011; de Jong et al, 2022).

The potential risk of too much proximity is that it can easily result in excluding other actors, or create biases and blind spots in collaboration (Boschma, 2005). Such lock-in effects can also be counterproductive for the added value of diversity in transdisciplinary collaboration, where heterogeneous and diverse perspectives and practices from participating actors is considered important for fruitful and transformative collaboration (Mauser et al, 2013; Flinders et al, 2016; Klenk and Meehan, 2017).

Our study has taken the existing hybrid research spaces of one university in one particular city as an empirical focus. This has several limitations. First, taking Rotterdam as common geographical background enabled the study of the effects of organisational and epistemic differences in the coordination between science and policy. However, a comparison with collaborations at other geographical scales, or in different cities, would have enabled us to analyse with more clarity the importance of the social, cultural and political values and relations in this particular urban setting (Malmberg and Maskell, 2006). On the other hand, this shared locality allowed the articulation of geographical differences at smaller scales, such as shared working space and intra-city travel time. For future research it would be worthwile to undertake studies in other local settings or with a comparative, multiscalar approach.

Second, the empirical focus on the spaces of interfacing in a limited timeframe also precluded an analysis of organisational adaptation, that is, the extent to which more proximity also results in institutional changes within the different organisations that could change the relations between science, policy and practice in a fundamental way. Future research could take the heterogeneous organisational processes that touch upon, flow into, and follow from the collaborative research work as a focal point.

Another limitation is that we took visibly institutionalised collaborations as relevant cases for our study, which by definition has excluded informal, less visible and otherwise subterranean relations between research and the ‘real world’ (de Sousa Santos, 2007; Stengers, 2018). It must also still be seen how easily the proximity frame can be used for less intensive, less localised or more ephemeral multi-actor collaborations. For example, future research could employ proximity to compare different types of bridging work, not only between science and policy but also ‘integration work’ more generally (Hoffmann et al, 2022).

Coordination work between science and policy remains ambiguous and a balancing act, and describing it in terms of proximities has allowed awareness of the political dimension: organising for proximity also implies stirring up dynamics of inclusion and exclusion. Organising more proximity with one target audience easily enlarges distance for another. This is an important aspect to take into account when it comes to organising the boundaries between science and policy: it is as much about organising proximity as it is about organising distance. Our recommendation to scholars, policymakers and other (knowledge) actors involved in such collaborations is to use a proximity approach as a discussion tool for the participatory design of strategies, to enable politically sensitive productive interactions. The latter can be specified in terms of goals, actors and ways of working, and the organisational and geographic dimensions of proximity can be employed as intruments to stimulate the required degrees of social and cognitive proximity, while mitigating the exclusive tendencies proximity inherently entails.

Funding

This work was supported by Erasmus University Rotterdam and the Municipality of Rotterdam.

Research ethics statement

Research ethics approval was sought from the Ethical Committee of the Department of Public Administration and Sociology (DPAS) of the Erasmus University Rotterdam. They granted formal approval for the research project ‘Evaluating Societal Impact’ on 24 February 2021 (reference 21-002).

Contributor statement

JS, HW and AB designed the study. JS and HW conducted data collection, analysis and interpretation, with contributions from AB. JS, HW and AB wrote the first draft. JS and HW wrote subsequent and final drafts, with comments from AB.

Data availability statement

The interview guideline, focus group format and coding book are available in the Open Science Framework repository at doi:10.17605/OSF.IO/UZ6VD. Primary interview data could not be shared because anonymity and restricted reuse were promised to respondents in the informed consent process.

Conflicts of interest

This scientific paper builds on the results from an organisational evaluation study that was funded by the two institutions that are also the main funders of the science-policy collaborations analysed in this paper.