Introduction

Data management is now a major challenge for research organisations. Vast quantities of born-digital research data are now being produced in a wide variety of forms and at a rapid rate in universities, creating the so-called “volume”, “variety” and “velocity” challenges of data [1], [2]. This “data deluge” generates a need to develop policies, infrastructures and services in institutions to manage data, with the aim of assisting researchers in creating, collecting, manipulating, analysing, transporting, storing and preserving datasets. In certain circumstances, there may also be a need to share data either amongst specific groups or openly. Despite its challenges, there is growing recognition that sharing data widely can create benefits, including allowing the verification of research outcomes, enabling contribution to wider data gathering scientific activities, and facilitating the reuse of data by others for subsequent research [3], [4]. Research funders, conscious of the need to encourage scientific good practice and to achieve greater value for the research they sponsor, widely encourage – indeed, increasingly require – particular standards of data management and sharing to be followed [5], [6]. Such requirements serve to emphasise the need for effective research data management within institutions.

As different approaches to research data management (RDM) are developing in universities and other research organisations, different stakeholders have become involved, including support services staff as well as Faculty themselves. University libraries have moved into this space and are increasingly seen as major contributors to RDM activity in general and in the design of research data services (RDS) in particular [7]–[14]. The study presented here makes a contribution to the ongoing discussion in the research and professional literature by reporting results of qualitative research consisting of interviews of library professionals in UK universities discussing their involvement in institutional RDM activity. It is designed to provide an in-depth insight into the thinking of library staff involved in RDM as they confront the wide range of challenges involved.

The paper begins with a brief overview of the research context, discussing in particular the concept of RDM and the literature on librarians' roles in it. It then goes on to outline the objectives of the research undertaken and the research methodology used. The results then presented provide perspectives on the relationship between the library and other institutional stakeholders involved in RDM, the sorts of RDM activities being carried out, the drivers for those activities, and the main factors that influence their shape. From the data presented, a model is constructed which attempts to capture the main aspects of an RDM programme in an institution.

Research Context

Research data management is defined by Whyte and Tedds as, “the organisation of data, from its entry to the research cycle through to the dissemination and archiving of valuable results” [15]. As Cox and Pinfield observe, RDM “consists of a number of different activities and processes associated with the data lifecycle, involving the design and creation of data, storage, security, preservation, retrieval, sharing, and reuse, all taking into account technical capabilities, ethical considerations, legal issues and governance frameworks” [8]. Such activities and processes are needed for a wide variety of different forms of data ranging from large-scale calculations derived from high-performance computing facilities, through results of scientific experiments, to audio recordings of interviews. RDM is therefore a highly complex set of activities involving an array of technical challenges as well as a large number of cultural, managerial, legal and policy issues.

The library's involvement in RDM has been discussed in the literature particularly in the last 5 to 6 years. Early contributions, such as those by Delserone [16], Henty [17] and Lewis [7], set out the case for library involvement. Later works, such as Corrall [18] and Cox, Verbaan and Sen [19] further discuss the range of possible roles. Lyon [20] identifies a number of opportunities for libraries but also major challenges in developing the capacity and capabilities to carry out RDM. Pryor [12] summarises these opportunities and challenges by talking about “the re-purposed librarian”. Procter, Halfpenny and Voss [21] place the role of the library in a wider institutional context, emphasising the need for libraries to work in partnership with IT services and academic staff.

Further studies have attempted to gather empirical evidence of the extent of actual library involvement in RDM. Tenopir and colleagues [10], [22], [23] report the results of a large-scale study of US libraries' involvement in research data services design and implementation, showing activity in technical infrastructure development as well as support and advisory services. Corrall, Kennan and Afzal [9] report results from a survey of university libraries in Australia, New Zealand, Ireland and the UK carried out in the first quarter of 2012, identifying widespread early activity albeit constrained by knowledge and skills gaps. They assess early work in these countries, particularly the UK, to be somewhat behind those of the US. Cox and Pinfeld's study [8] of UK libraries presents survey results from late 2012 showing progress in the UK, with policy development involving the library in particular proceeding rapidly and a high strategic priority being given to some RDM activities especially around training and advisory services.

Empirical studies such as these have, however, to date largely been quantitative in their approach. Whilst they have identified a range of important developments, they have perhaps also highlighted the need to gain a richer and more nuanced picture of ongoing developments. The current paper goes some way to achieving that by making an early qualitative contribution, to date an approach largely absent from the literature. It complements work recently published by the authors based on the same dataset focused specifically on analysing the challenges of RDM within the framework of the ‘Wicked Problem’ concept (“unique, complex problems which are defined differently by different stakeholders making them particularly intractable”) [24]. The current study also follows on from quantitative work by Cox and Pinfield [8], allowing issues identified there to be explored in more depth.

Complementing empirical studies, a number of researchers have generated models of various aspects of Research Data Management. For example, Jones, Pryor and Whyte delineate a “Components of RDM support services” model [25], organised within a simple research life cycle framework (further discussed by Jones [26]). There is an overarching “Policy, strategy and business case”. Then there are a number of components relating to different stages of research, namely: support for data management planning; managing active data; data selection and handover; sharing and preserving data (including data repositories and catalogues). “Guidance, training and support” covers support services to researchers throughout the process. The model gives detail about the types of services that could be offered around RDM, and one of its strengths is its link to a simple data lifecycle concept. It could be seen as a successor to the well-known Digital Curation Centre (DCC) Lifecycle model, which is rather complex and focuses more on data curation than on researcher support. Complementing the Service component model, Whyte proposes a process by which services might be brought into being, through a six-step pathway of envisioning, initiating, discovering, designing, implementing and evaluating [27]. The author sets out the tools that have been developed by the community to support each step of the process. He acknowledges some of the challenges, particularly in the diversity of research communities, but does not provide a detailed consideration of any constraints on the service delivery side. Mayernik et al provide an outline of a “Data Conservancy” model of RDS based on work at Johns Hopkins University in particular. This covers both technical components (such as software and infrastructure) and organisational components (such as policies and funding strategies) [28].

Good practice has also begun to emerge as institutions have set out to develop RDS and have published lessons learned based on real-world experience. Experiences in institutions such as Edinburgh [29] and Oxford [30] in the UK, and Purdue [31] and Johns Hopkins [32] in the US, have highlighted some important issues at institutional level. Differences in national approaches between Australia and US have also been usefully explored [33]. However, even the advanced institutions identified in these publications offer only hints of what a fully developed RDS may look like. At the time the interviews reported in this paper were carried out, many institutions were clearly only taking their first steps towards finding a pathway.

The research had four main objectives designed to analyse the library's contribution to RDM within the wider institutional context:

1. To examine roles and relationships involved in RDM
2. To identify the main components of RDM programmes
3. To evaluate the main drivers for RDM activities
4. To analyse the key factors influencing the shape of RDM developments

In this way, the research was designed to yield a rich understanding of current institution-wide RDM developments as contributed to and perceived by one of the major stakeholder groups – the library.

Methodology

The data presented here derives from 26 semi-structured interviews of library practitioners from different institutions in the UK. These interviews were the second phase of a project which began with an online survey of UK academic libraries reported by Cox and Pinfield [8]. Respondents to the survey were asked to volunteer to take part in detailed follow-up interviews as part of their responses, and therefore interviewees represented a self-selected group of practitioners, probably engaged with RDM to a greater degree than the community as a whole (although a number of participants were only just beginning to address RDM issues when interviewed). Participants came from a number of different sorts of institutions (research-led and teaching-led institutions, large universities and small institutes) and were either senior library managers with a strategic overview or middle-level managers with direct responsibility for RDM.

The approach adopted for the research was given approval under the University of Sheffield Information School ethics approval process as overseen by the University of Sheffield Research Ethics Committee. The University Research Ethics Committee monitors the application and delivery of the University's Ethics Review Procedure across the University. The process as carried out was based on the principles of voluntary contribution, informed consent and anonymised reporting. Participants, who had volunteered to take part by providing their contact details in the previous online survey reported by Cox and Pinfield [8], were sent Information and Consent documents in advance of the interviews which explained details of the research approach and their part in it. Participants were given the opportunity to ask questions about these via email before the interviews were scheduled. At the beginning of the formal telephone interviews, each participant was given the opportunity to ask questions verbally regarding their participation and asked to confirm their consent. Verbal consent was deemed appropriate as interviews took the form of pre-arranged telephone conversations with self-selecting volunteer participants. 25 of the interviews were conducted by telephone, with one face-to-face, at the request of the particular participant.

The interviews covered three main areas: firstly, the current state of RDM activity in the institution; secondly, the skills needed to carry out RDM services; thirdly, the story of how RDM policy and services have developed in the institution. Interviewees were encouraged to talk in detail about their experience of the issues involved, including the library's role in delivering RDM and in developing RDM-related strategies and policies, with interviews lasting 40 minutes on average. Following piloting, interviews took place between March and June 2013. Interviews were recorded and then fully transcribed for analysis.

The interview transcripts were analysed in NVivo following a thematic analysis approach [34]. All three authors initially familiarised themselves with the data with a thorough reading of the transcripts, involving memo writing. Initial codes were then identified based on key features identified in the data, including “semantic content or latent” features [34]. Codes were applied to the transcripts in NVivo, and reviewed and refined on a number of occasions by the authors following review. After initial coding of all of the transcripts, major themes were identified from the data, with code structures refined and thematic maps developed to reflect these. Codes and themes were reviewed and again refined upon further analysis of the data. The data was then used to construct a model of an RDM programme in the institution from a library perspective which was designed to capture the key themes highlighted in the interviews (presented below).

Results

Factors Influencing RDM Developments

As well as describing the different aspects of RDM activity and the key drivers, participants also identified a large number of factors which shape an institutional RDM programme and the library's involvement in it. These might be called ‘Influencing Factors’ since they are intervening conditions which may affect an institutional RDM programme in a variety of complex ways including either facilitating or constraining action. Key Influencing Factors emerging from this research were:

• Acceptance
• Cultures
• Demand
• Incentives
• Roles
• Governance
• Politics
• Resources
• Projects
• Skills
• Communications
• Context

The first of these, the level of acceptance and prioritisation of RDM as a coherent agenda, is related to the jurisdiction issue above. It was clear from participants' comments that the RDM agenda was not universally accepted in their institutions. Some mentioned scepticism amongst colleagues from other support departments, particularly IT services whose attention was often focused on questions of storage and security only. Others highlighted a lack of engagement from researchers. In either case, it is clear that this lack of acceptance would have a major impact on the way in which an RDM agenda was formed and implemented. In other cases, the RDM agenda may have been recognised but was given lower priority than other key issues. For example, some participants reported difficulties in taking forward the RDM agenda because of the priority given to the research excellence framework (REF) exercise by research support staff and senior academic managers such as pro-vice chancellors.

Added to this, perhaps one of the most significant challenges in implementing institution-wide initiatives was varying cultures and consequent differing working practices. Within any single institution there were seen to be a large number of different cultures across different professions and academic disciplines. Any institution-wide RDM programme, it was recognised, needs to take such varying cultures and practices into account in the way it is designed and implemented. Many of the participants in this study were clearly aware of this issue and mentioned in particular disciplinary differences. Differences manifested themselves in the variety of forms of data being generated and ways in which it was analysed. There were technical differences in areas such as metadata standards and interoperability protocols, but also in cultures around sharing and reuse. However, in many cases, participants were clearly only just beginning to take account of the implications of this for institution-wide RDM programmes. Nevertheless, despite the lack of detailed solutions discussed by participants, the importance of the issues of disciplinary differences should not be underestimated. In many respects, it colours all of the other Influencing Factors.

One of the major challenges implicit in the remarks of many interviewees was that the drivers relating to RDM in many cases operate at an institutional level and are not created by user-level demand in the academic community, regardless of discipline. Apart from data storage, it was clear from responses that demand for RDM-related services amongst users themselves was in fact limited or absent. Instead, participants saw the challenge as lying in persuading users to recognise the importance of certain approaches and adopt particular practices associated with RDM, rather than in responding to user demand. A clear understanding of where user demand does exist and where there is a perceived need for particular services (with or without explicit demand) is certainly therefore an important factor that will shape ongoing RDM activity.

Where there may not be extensive user demand it was seen as essential to identify the key factors that determine behaviours, particularly in terms of incentives. Developing a policy, for example, which includes clear incentives for encouraging desired behaviours, and also clear sanctions to discourage undesired behaviours, was essential. There was clearly uncertainty amongst participants about what such incentives or sanctions should look like at an institutional level in relation to RDM, although the role of the requirements of external funders was seen by many as important in shaping institutional approaches.

Linked to this, a clear understanding of the roles and responsibilities of different stakeholders in institutions emerged as essential in determining the shape and momentum of RDM activities. Just identifying all the stakeholders and their potential roles was problematical. This applied to roles of organisational units, such as the library or IT service, as well as individuals, such as principal investigators or heads of academic departments. Participants reported some uncertainty around all of these in their institutions. There was ongoing uncertainty, for example, in institutions about whether the library was seen as a ‘natural’ place to go for research data management services, although to balance this there was reported to be a high level of acceptance of the library being involved at least in initial stages of RDM programmes around policy development.

Institutional approaches to governance were also recognised to be very important, particularly in terms of decision-making. Universities are often characterised by governance complexity and ambiguity, meaning that decision-making can be a lengthy and opaque process with somewhat indeterminate outcomes. Implementing decisions widely and consistently therefore creates major challenges. Many participants expressed frustration at the delays inherent within University decision-making processes and also of the fact that policy decisions at institutional level did not necessarily result in compliance across the organisation as a whole.

Related to this, the power dynamics within the institution particularly around personal relationships of senior managers were often identified as being important by participants and labelled as “politics”. The relationship between the library director and other senior members of staff, particularly the Pro-Vice-Chancellor for research, Director of research support services and Director of IT services were seen as crucial in taking an RDM programme forward. Securing the attention and time of such individuals and their staff were seen as essential and was a consistent feature of the interviews. As RDM activities matured, relationships between staff in different departments below senior team level became more important and it was here that some participants were beginning to encounter challenges.

Resources available for an RDM programme were seen as crucial. Whilst it was clear that there was still a great deal of uncertainty about the funding and resources required to carry out RDM activities within the institution, there was scepticism about the extent to which new resources would be identified for such activities. Although in some instances business cases for new funding were being developed, most libraries were clearly operating within an environment where efficiency was given a very high priority and gaining additional resources except for fixed-term projects was seen as difficult. Where new staff were required by the library or organisational restructuring was seen as necessary, this would often have to be achieved through repurposing existing resources. Such an atmosphere clearly constrained the scale and scope of RDM activity and services envisaged by many participants.

In many cases, as has been observed, RDM programmes were being focused on fixed-term project activity aiming to deliver specified outputs including policy and guideline documentation and pilot services. These projects were often funded by external agencies, such as Jisc, although a number of participants mentioned internally-funded projects usually also with fixed-term funding. As well as being efficient, these activities were often seen as a vehicle for exploring issues in a manageable way, attempting to tame the scale and complexity of the problem often, for example, with a particular disciplinary focus. It was, however, clear that in the case of many institutions there was still significant uncertainty about next steps following the completion of such projects. In a small number of cases programmes of activity were being developed but were in the early stages. Moving from projects to services, involving scaling up pilots and embedding policies in everyday practices, emerged as a major challenge.

Early activity such as projects often highlighted major skills gaps in organisations, including libraries. Skills and expertise were seen as a major influencing factor on an RDM programme. There was, however, uncertainty amongst participants about the skills required to carry out RDM activities. At this early stage, many libraries had appointed project management staff but were in the process of identifying other areas of expertise necessary to carry out RDM on an ongoing basis. Particular areas identified included advocacy and liaison skills, training skills, as well as technical skills. A large number of participants identified the importance of re-skilling existing staff to be able to operate in the RDM space. However, in some cases, participants reported that they were planning to appoint new staff but this would be limited to only one person or a small number of people (often because of resource constraints). Such constraints were obviously impacting significantly on the shape of the RDM programme being undertaken.

Whether carried out by a new RDM Officer or by other staff within the library (such as subject librarians), communications, and particularly advocacy, were seen as an essential part of any library involvement in RDM. Developing a communication and consultation process in order to determine the content of policies and guidelines and to shape the design of services was seen as essential by participants. The library had an important role in advocacy – articulating the importance of RDM for the institution and its various stakeholders. This could often be achieved by the library its existing channels of communication with particular subject communities established by subject librarians or research-support staff. In addition, it became apparent from interviews that communication within the library about RDM was also important. On occasions, it was clear that the views of senior library staff including the director in relation to research data management may have been out of synchrony with those of other staff. In some cases, following initial negotiation with partners in the University, library directors were devolving RDM activity to their own staff without there being sufficient clarity about what was now expected. Several such staff were participants in the interviews and were clearly going through a process of identifying what next steps were possible and which should be prioritised.

All of these factors affecting the shape of RDM programmes in individual institutions were determined in part by the setting and context within which they were taking place. Research-led institutions were clearly structuring their activities differently from teaching-led institutions, for example. Disciplinary coverage within institutions, varying from multi-disciplinary universities to subject-specific institutes, was also an important factor. It was clear, as one participant put it, that there is a “need to have an understanding of the landscape of your institution” and that institutional setting and context were important influencing factors in the development of an RDM programme.

Discussion

This analysis reveals a complex picture. While libraries saw that issues around RDM were something to which it was important to respond, their role varied markedly. Any response would be inherently collaborative, but it was clear that solutions developed would vary across different institutions. This is partly because of the existence of a number of different drivers and a large number of inter-related factors influencing how services might be created. It has been suggested that UK institutions have been rather slow to respond to RDM by comparison with the US in particular [9]. The picture from the data in this research reinforces the sense of rather cautious first steps being made, not because of poor leadership (if anything, libraries are leading the way) but constrained by a lack of direction at institutional level. This could have a number of explanations, not least of which is resource limitations. To balance this, however, there was a clear view expressed by participants that the library should be involved in a significant and sustained way in RDM.

Combining the findings of this research with issues identified in previous quantitative work [8], a tentative model of an RDM programme within an academic institution has been constructed (Fig. 1). This model helps to clarify the different issues involved in the RDM challenge identifying various layers of activity, multiple stakeholders and drivers, and a large number of factors influencing the implementation of any programme. The model has been created with the library's perspective in mind but also applies more widely across the institution.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. A library-oriented model of institutional RDM.

https://doi.org/10.1371/journal.pone.0114734.g001

The model is intended to address the ‘who?’, ‘what?’, ‘why?’, and ‘how?’ of RDM, particularly in relation to the library's involvement. The ‘Stakeholders’ in the model address the question, ‘Who is involved in institutional RDM?’. They constitute the main actors in the phenomenon being analysed. The ‘RDM Programme Components’ in the model address the question, ‘What does an RDM programme in an institution consist of?’. They constitute the main elements of the phenomenon itself. The ‘Institutional Drivers’ address the question, ‘Why might a programme be carried out?’. They constitute the main causal factors of the phenomenon. The ‘Influencing Factors’ address the question, ‘How will the programme be shaped?’. They constitute the main intervening conditions that affect the phenomenon.

In the model, the different components of an institutional RDM Programme (strategies, policies, guidelines, processes, technologies and services) are shown as logical layers of related activity, from high-level strategic planning to on-the-ground service deployment. Between them, policies and guidelines constitute what might be called a ‘Regulatory’ layer in the model which may often be supported with agreed monitoring mechanisms to measure compliance. Processes and technologies between them constitute a ‘Systems’ layer in the model – systems which are socio-technical in nature.

The different components of an RDM Programme are shaped by a set of Drivers which have been identified by this research: storage, security, preservation, compliance, quality, sharing and jurisdiction. The Drivers identified in the model are not presented in any kind of priority order; notions of priority clearly differ amongst different stakeholders in different institutions and at different times. However, they do move from ‘harder’ technical challenges to ‘softer’ policy and managerial issues. The decision has, however, been made not to distinguish between drivers that are ‘external’ or ‘internal’ to the institution. This study suggests that all of the drivers presented here have external and internal aspects to them with a complex set of issues and influences at play for each driver both within a single institution and beyond.

The RDM programme components and drivers interact with a set of Stakeholders, all of whom have a role in RDM within the institution. Typically, as has been seen, the stakeholders include support staff – the library, IT services, the research support office, other support services (including records management and legal advice functions) – and academic staff – senior university managers (normally led by the Pro-Vice-Chancellor for research, or equivalent) and researchers in academic departments. Amongst academic staff there are a very large number of different researchers with (depending on the institution) different disciplinary approaches, represented in a separate layer in the model. Any institution-wide programme needs to take these disciplinary differences into account. The precise roles of the stakeholders and the relationships between them will, of course, vary between institutions, although it is clear from the research presented here that there are significant commonalities between institutions, including the identity of the players themselves and the particular perspectives they tend to take on RDM.

The implementation of an RDM Programme is affected by a complex set of Influencing Factors which impact on and are impacted by the Programme, either facilitating or constraining action, and influencing its character and direction. They will influence a programme in different ways at different stages (and therefore are not presented in any prioritised order). Each one is in its own right an area of considerable complexity, with additional complexity derived from the interrelationships between different Influencing Factors.

The model aligns well with previous work such as Jones, Pryor and Whyte [25], though operating at a different level. In terms of service components, the current model is at a higher level of generalisation: it does not deal with the specifics of particular types of technologies and services, such as data catalogues, which are more fully developed by Jones et al [25]. In a sense the components set out by their model could be substituted for the “What” part of the model above. At that level it should be clearer how the different professional services' roles might be mapped to the different types of user services, e.g. Active data storage and security – IT; data catalogue – library; Data management planning – Research administration; Guidance, training and support – all. However, the model presented here deals more with the complexity of the underlying drivers and influencing factors that could shape which types of specific service are developed and what they might look like – with a strong acknowledgment that there could be many different outcomes. There are again connections to Whyte's pathway to an RDS [27], although his work focuses on a process of designing a service, as opposed to representing the forces at work which could drive a programme in a particular direction or block easy progress. While it is clearly not the intention of the authors of these previous studies to say “one size fits all”, their approach is to delineate current best practice thinking. The current model makes it clearer why in practice very different patterns of support service might emerge, or indeed none at all, because it retains a complex sense of the underlying drivers and acknowledges the constraints on putting services in place.

In any particular RDM institutional programme it is reasonable to assume that successful outcomes will be achieved if all of the central components identified (strategies, policies, etc) are implemented by the appropriate stakeholders, in line with the drivers, and taking into account the relevant influencing factors. However, there is no single development pathway. Many institutions are evidently formulating and documenting policies as an initial stage and following this up with the creation of more detailed guidelines as processes and technologies begin to be put in place. There is activity in the areas of technology development and piloting of services. At the same time, work to identify and allocate required resources, clarify roles and responsibilities, and enhance necessary skills development are under way. Commonly, this is being done through project activity, which is also enabling some advocacy to take place, but it is clear that the scaling and operationalisation of activity remains challenging. Moreover, ensuring that all the different components of the programme (policies, guidelines, processes etc) are all developing at a consistent rate and with appropriate levels of compatibility and integration is also challenging. Managing the different strands of activity being carried out by different actors (the library, IT services, academic departments etc) together to form a coherent institutional approach without unnecessary duplication or incompatibility is becoming a concern for many.

This model serves to encapsulate reported RDM activity within institutions, and it may also act as a diagnostic tool. If an RDM programme is stalled, for example, it may help to analyse progress against the model in order to identify areas where little or no activity has been focused or where there is a significant constraining influence. Approaches can then be developed to address these issues.

Of course, discussion on the questions, ‘who?’, ‘what?’, ‘why?’, and ‘how?’ begs the question, ‘when?’. This is not been built into the model since the participants in the research presented in this study were very uncertain about timescales. Whilst quantitative data reported by Cox and Pinfield indicates that libraries expect to have made significant progress in RDM “in the next three years” [8], this may often be a statement of aspiration rather a summary of specific plans. The fact that it is “early days” and that, apart from policy development, libraries in particular and institutions in general are still “feeling their way” with RDM means that ‘when?’ is still very much an open question.

One under-developed aspect of the model is the treatment of the way that any initiative needs to be adaptive to the diverse disciplinary cultures of research communities [27]. For example, it is well understood that some disciplines already have a deeply ingrained culture of sharing data openly (physics) or reuse of secondary data (economics); others have well-developed data management practices, driven by the use of personal data (health sciences); still others, such as in some of the arts and humanities, rarely even use the term “data”. Such issues are incorporated in but are not central to the current model. This is partly a function of the data used to produce it: namely, librarians in the early days of planning RDM activities, before intense engagement with research communities around initial services. Although participants were conscious of the importance of disciplinary differences, the ways in which these should be reflected in the detail of institutional services was still often unclear. Data based on the perspective of other stakeholders, such as researchers, but also other professional services, might also lead to the model being refashioned. As has been stressed, the data reported here are the library perspective on the issues: this viewpoint is important in itself, but it is a partial perspective. So the model should be seen as a first tentative formulation based on one dataset – but one that does accurately capture the perspective of libraries on the early days of developing RDM.

At the same time, it seems plausible that elements of the model may be generalised to apply to other institutional information-related initiatives in higher education organisations. Whilst the specific drivers and stakeholders will differ, the programme components and influencing factors are likely to apply in other cases, although levels of complexity may differ depending on the specific initiative. Initial analysis suggests that it would, for example, be applicable to the development of an institutional open access programme in this form. Here the development of a strategic approach to open access accompanied by appropriate policies and guidelines, development of relevant technologies and processes, and provision of supporting services might be seen to be influenced by similar factors to RDM. Furthermore, other institutional initiatives outside of the information field also arguably follow a similar pattern. Testing the utility of this model for other developments in HE, both information-related and other initiatives, may be a useful area of further research.

Conclusion

Research data management is a complex issue involving multiple activities carried out by various actors addressing a range of drivers and influenced by a large set of factors. The analysis and modelling of developments presented in this study contribute to the understanding of the ways in which institutions are addressing the problem by presenting in detail the perspective of one major stakeholder group. Whilst the analysis has focused on the activities of libraries in particular, it illustrates more generally how different actors are adapting their roles to participate in emergent RDM programmes. However, major uncertainties remain in how the various stakeholders relate to each other, where strategic priorities lie, and how socio-technical systems should best be designed to deliver value to the organisation in particular and research community in general. Library activity, currently concentrated in areas such as advocacy and policy development, and moving into new areas including the support functions and creation of new systems, still has an important element of provisionality about it. As RDM matures in universities, further quantitative and qualitative work will be needed to understand the shape of activities and the roles of different actors in order to inform ongoing development.