Digital Preservation and Stewardship Committee

Research Data Working Group

Interim Report and

Proposal for the

Atlantic Research Data Repository (ARDR)

   

Michael Beazley Suzanne van den Hoogen

Karen Keiller Mark Leggott Mike Nason 

Maggie Neilson Kathryn Reddy 

 May 12, 2015 

 

 

Report Contents

EXECUTIVE SUMMARY BACKGROUND

History of the DPSC Overview of National/Regional/Institutional Cyberinfrastructure

SURVEY Introduction Results

Collection Services User Services Access Services Preservation Services

Gaps and Requirements Summary and Discussion RECOMMENDATIONS AND NEXT STEPS Introduction Survey Recommendations Research Data Management Infrastructure Research Data Management Planning Tool

Regional Research Data Storage Service CAUL/CBUA RDM Team Governance and Administration Sustainability

APPENDICES Appendix A: Acronyms and Glossary Appendix B: Bibliography Appendix C: CAUL/CBUA Research Data Management Survey Instrument Appendix D: CAUL/CBUA Survey Infographic Appendix E: Comprehensive Brief on Research Data Management Policies

2 | Page

 

 

EXECUTIVE SUMMARY  The interim report of the CAUL/CBUA Research Data Working Group reflects the discussions of the Working Group for the last 18 months, which centered around the emerging interest in supporting research data management. A key part of this effort was the completion of a research data management survey of CAUL/CBUA members, providing a baseline for the discussion of possible next steps for the consortium and its members. This was augmented with updates on efforts of national initiatives and reports of activities from member institutions. These discussions led to a number of recommendations, which are summarized below and explored in more detail later in the document.

1. Research Data Management Survey a. That the survey results be posted in full on the CAUL/CBUA website. b. That the survey be completed on an annual or biennial basis. c. The survey questions be evaluated and updated as necessary for each survey

cycle. d. Employ a survey tool (such as Survey Monkey) which can provide a long term

framework for maintaining the survey. e. Offer a response category labeled “in progress” to allow institutions to indicate

services that are in development, but not yet fully realized. f. Provide more opportunities for comments within the survey, to allow respondents

to clarify or elaborate on answers to specific questions. 2. Research Data Management Planning Tool

a. CAUL/CBUA endorse a RDMP tool, with an eye to the RDMP tool currently under development at the University of Alberta, CARL and Compute Canada.

b. Locally deployed tools be able to export plans to the national repository, contributing to a national inventory of RDMPs.

3. Regional Research Data Storage Service a. CAUL/CBUA support the creation of the ARDR service, which would function on an

“all-in” cost-sharing model and would provide data storage and services to all member institutions.

b. The ARDR system provide a two-tiered service model, offering a basic service package as well as a value-added service package.

c. The hardware infrastructure (i.e. servers, storage drives, etc.) be hosted in four locations: UPEI, UNB, Dalhousie, and MUN.

3 | Page

 

 

4. CAUL/CBUA RDM Team a. Establish a CAUL/CBUA Research Data Management Team. b. Establish a support network akin to the current Data Liberation Initiative (DLI)

model. 5. ARDR Governance and Administration

a. Adopt a model policy that CAUL/CBUA would endorse for the oversight of ARDR, and individual institutions would use as the basis for local policies.

b. CAUL/CBUA institutions adopt a minimal RDM preservation commitment that says: “The University will steward the data for [Project X] for as long as is needed.”

c. ARDR services and policies be crafted with an awareness of national RDM initiatives.

6. ARDR Sustainability a. Propose an initial 3-year financial investment from all CAUL/CBUA members to

get the project started. b. CAUL/CBUA look for grant opportunities (ACOA, TC3+, etc.) to assist with the

start-up costs.

4 | Page

 

 

BACKGROUND History of the DPSC

Committee Formation The CAUL/CBUA Digital Preservation and Stewardship Committee was first proposed at a Board of Directors meeting in October 2009. The terms of reference for the Committee were developed in early 2010 through meetings with Mark Leggott, Donna Bourne-Tyson, Lynne Murphy, Tanja Harrison, and Bruno Gnassi. The initial membership of the DPSC included: Donna Bourne-Tyson (Chair), Tanja Harrison, Karen Keiller, Gillian Byrne, Mark Leggott, Dawn Hooper, Alain Roberge, and Slavko Manojlovich. Priorities In May 2012, the DPSC recommended adopting a working-group approach to focus on specific activities, as forwarded by the Directors. At that time the following priorities were approved:

1. Develop and adopt principles, guidelines and an infrastructure capable of sustaining digital/data preservation and stewardship.

2. Develop initiatives to test and advance these principles and guidelines and build on infrastructure and capacities that may already exist as regards to digital/data preservation and stewardship.

3. Organize training and promote cooperation amongst CAUL/CBUA members to advance digital/data preservation and stewardship.

4. Foster a culture within CAUL/CBUA committed to the sound management and advancement of digital/data preservation and stewardship.

5. Liaise with other regional, national, and international bodies on behalf of CAUL/CBUA. First Digital Preservation Survey In November 2012, the DPSC compiled a list of priorities. One of the initial priorities was to distribute a survey “to assess the state of CAUL/CBUA members' current digital preservation (DP) activities” and further facilitate the work of the DPSC. The results of this survey were presented by Marc Truitt (DPSC Chair) at the May 2013 Board of Directors meeting. Some of the core findings included the following:

● For some members, responsibility for digital preservation exists outside of the library's core area of responsibility. Examples include the following: academic computing units (or IT departments), archives, and records management.

● Few institutions have written digital preservation policies and procedures. ● Common among institutions appear to be audiovisual files, PDFs, digital image files,

word-processing files, licensed e-journal files, and institutional records. ● Dealing with obsolete external media is a common experience among members, especially

with regards to reading files from obsolete media such as 5.25 disks. It was noted that

5 | Page

 

 

smaller institutions might benefit from the sharing of expertise and knowledge of other libraries who have experience with file migration and playback. This is an opportunity for CAUL/CBUA to play a role in facilitating the sharing of information between institutions on how playback devices might be acquired, set-up, and used appropriately.

● Online and disk-based (external media) appear to be common backup solutions employed by reporting institutions.

● At least one institution is interested in achieving the status of Trusted Digital Repository; either by its own efforts or through a CAUL/CBUA-sponsored initiative.

● Dalhousie, Memorial, UNB, and UPEI have established institutional programs for digital preservation.

● Given the diverse states of CAUL/CBUA members' individual digital preservation activities, there is a place for CAUL/CBUA to advance preservation activities regionally.

Establishment of Working Groups In late 2013, Mark Leggott assumed the role of Chair for the DPSC. Membership included Lou Duggan, Mark Leggott, Erik Moore, Karen Keillor, Nicole Dixon, Creighton Barrett, David Mawhinney, and Roger Gillis. Discussions led to the development of five working groups, each with a lead from the DPSC, whose responsibility was to advance the goals of the group, develop group membership, and encourage a culture of learning through working group discussions. The working groups and leads began, as follows:

1. Digitization and Preservation Policies: Nicole Dixon (CBU) 2. Processing of Obsolete Media: David Mawhinney (MtA) 3. Research Data Working Group (Formerly Regional Cloud Storage Pilot): Mark Leggott

(UPEI) 4. Regional TDR/TRAC Framework: Creighton Barrett (DAL) and Erik Moore (UNB) 5. Digitization/Preservation of Government Documents: Roger Gillis (MSVU)

The Digitization and Preservation Policies Working Group has since been integrated with the Regional TDR/TRAC Framework Working Group. The Processing of Obsolete Media Working Group released a report in 2013 and has since been disbanded. In early 2015 the issue of the preservation of Government Documents was also assumed by the CAUL/CBUA Collections Committee, so areas of overlap will be considered as this group advances. As a result, there are currently two active working groups in the DPSC. Research Data Working Group In February 2014, the Research Data Working Group began meeting on a regular basis under the following terms:

1. Document existing services and resources at CAUL/CBUA member institutions directed towards the stewardship of research data, and consider the development of a common

6 | Page

 

 

instrument for determining faculty and graduate student interests around research data stewardship.

2. Review options and make recommendations for providing research data stewardship services for members, including expertise, storage, processing, training, and including the role of Liaison Librarians in this effort.

3. Ensure that recommendations facilitate institutional, regional, national, and international efforts to steward research data and that they intersect with existing and emerging mandates at all levels.

4. Provide information regarding opportunities for funding, including possible partnerships with CARL, CRKN, RDC (Research Data Canada) and other national and international initiatives.

This Interim Report is the outcome of these meetings and includes the working group’s recommendations on responding to the need for research data management.

   

7 | Page

 

 

BACKGROUND Overview of National/Regional/Institutional Cyberinfrastructure Note: This is a brief summary of the current state of Canadian research data stewardship and is enhanced with Appendix C: Comprehensive Brief on Research Data Management Policies created for Canada’s Tri-Council agencies by Kathleen Shearer, which provides a comprehensive review of approaches to data stewardship internationally and nationally.

The Canadian landscape for research data stewardship is in its infancy in comparison with other countries. In many respects, the U.S., UK, and Australia define the current state-of-the-art in research data management. In Canada, individual institutions are starting to develop policy-driven approaches to research data management, such as Simon Fraser University, the University of British Columbia, the University of Alberta, and the University of PEI. These efforts are an attempt to lay the groundwork for institutional services that represent best-practices, in anticipation of the introduction of a research data management mandate for researchers receiving Tri-Council funding. This effort has been highlighted by Canada's Action Plan on Open Government 2014-16 , which calls for the development and adoption of policies, guidelines and 1

tools to support effective stewardship of scientific data. In addition to these efforts there are two substantial national initiatives underway:

1. Leadership Council for National Infrastructure, Pilot Project ● A multi-jurisdictional effort with primary oversight by Research Data Canada (now

funded by Canarie) and participation from CUCCIO, CARL, Compute Canada and a number of nationally-funded research projects, including CBRAIN, CADC, and IPY.

● The Pilot Project group is working to develop an approach to building national cyberinfrastructure for the stewardship of research outputs.

2. CARL Portage Project ● A CARL-initiated project designed to identify current institutional best-practices and a

pilot framework in which interested parties can determine service and resource options for a national approach. This effort was originally referred to as the CARL ARC project, based on the first meeting location in Ottawa. As the discussion proceeded the project assumed the name Portage.

Early in 2015, the CARL ARC project team submitted a document to CARL Directors, including a set of recommendations for a national RDM support service and a business model for sustaining the initiative. The recommendations included in this document reflect the Working Group’s assumptions that the CARL Portage project will proceed as outlined in this proposal, and that a regional CAUL/CBUA approach to RDM needs to consider the broader national efforts and ways to best integrate with those efforts. [Note: A copy of the CARL-approved Portage proposal will be

1 Government of Canada. (2014). Canada’s Action Plan on Open Government 2014-16. http://open.canada.ca/en/content/canadas-action-plan-open-government-2014-16

8 | Page

 

 

either included in this report or distributed to CAUL/CBUA Director’s when it is available.] Excerpt from the Portage Brief : 2

The aim of the Portage network is to pool and expand existing expertise, services and infrastructure so that all academic researchers in Canada will have access to the support they need for research data management. The Portage network will have two major components:

1. A library-based distributed centre of expertise for research data management; and 2. A national preservation and discovery system for research data that will evolve and expand

over time. Distributed Centre of Expertise RDM requires specialized knowledge and expertise, which many researchers do not have. The Portage centre of expertise will provide access to a comprehensive set of resources that point users to the most up-to-date, relevant and trusted sources about RDM. In addition, Portage will host a national web-based tool, to launch in early 2015 that will assist Canadian researchers in developing data management plans. Portage will also act as a forum for sharing expertise across the country in order to build institutional capacity. Areas of expertise will include: privacy, security, and confidentiality; skills and training; data management plans, data discovery, data curation and preservation. National Preservation and Discovery System Advice and support for researchers must be accompanied by viable technical solutions. To that end, Portage has also been working on a project to connect the various infrastructure and service components needed for a national preservation and discovery network. The project is being undertaken in close collaboration with Compute Canada, Research Data Canada, and some of the domain data centres to ensure that it will be both inclusive and interoperable. The project will soon begin to ingest data into two sites that will provide long term preservation services. Once any problems have been addressed and workflows have been stabilized, the network will expand to include other repositories. The ultimate aim is to enable all interested universities to participate, whether or not they have their own local infrastructure, by coordinating shared repositories and services under a cost model that recognizes varying institutional investments and needs.

2CARL. (2014, December 22). Portage: Supporting Canadian innovation through shared expertise and stewardship of research data. Retrieved from http://www.carl-abrc.ca/uploads/SCC/Portage-External-2-Dec-22-2014.pdf

9 | Page

 

 

SURVEY Introduction This section addresses the first priority for the RDWG: indication of gaps and requirements in research data stewardship within the CAUL/CBUA member institutions. In March 2014, DPSC RDWG members determined a survey would be the best tool for collecting reliable data and staff input from across the CAUL/CBUA community. A survey distributed by CARL was adapted with permission and distributed to Library Directors on April 13, 2014. CAUL/CBUA Directors were also asked to forward the survey to the appropriate staff within their institutions. Libraries who had previously responded to the CARL survey were encouraged to forward their results directly to the DPSC Chair, Mark Leggott. The survey focused on four aspects of research data services, with ten questions in each section:

● Collections Services ● User Services ● Access Services ● Preservation Services

Individual institutions were given four weeks to respond. Additionally, a reminder was sent to encourage participation in the survey. The survey met our goal of obtaining regional representation, with 16 out of 17 institutions responding. Survey administration and data collection was compiled by the DPSC RDWG. Following a meeting on June 3, 2014 at the annual Atlantic Provinces Library Association (APLA) Conference, a need to confirm survey results with respondents was discovered. Respondents were contacted for clarification where needed. Following the February 2015 Directors meeting, members were given additional time to update their survey response. The survey proved to be an effective tool which revealed the discrepancies within the CAUL/CBUA environment related to research data stewardship and preservation.

10 | Page

 

 

SURVEY Results Survey results indicate that there are significant gaps in research data stewardship across the consortia. 3

Collection Services

Collection services were described in the survey as “activities that specifically support the development, acquisition, management, description, and discovery of a collection of research data files”. Some key observations from the data are as follows:

● 68.8% of CAUL/CBUA respondents have a subscription with data providers such as DLI, DMTI, or ICPSR.

● 25% of respondents (DAL, MUN, SMU, and UNBSJ) supplement their data services with documentation such as user’s guides, data dictionaries, and variable lists.

● 25% of institutions (DAL, MUN, SMU, and UPEI) have metadata librarians or other specialists who advise on standards for content and technical metadata.

● DAL, MUN and UPEI are leaders in collection services and data. They are the only institutions polled who:

○ have a dedicated budget to purchase data files outside of subscription services; ○ maintain a collection of data files from local researchers; ○ maintain the infrastructure to manage local data file collection; and ○ produce standards-based metadata for research data.

● DAL is the only institution that is a member of a standards body for research data or metadata.

● DAL and UPEI are the only institutions with a written Collection Policy for research data.

User Services User services were described in the survey as “activities that focus on supporting user communities by identifying their data needs, assisting them in preparing data management plans, selecting metadata standards and best practices, identifying existing data sources, and retrieving, manipulating, and transforming data”. Some key observations from the data are as follows:

● 68.8% of CAUL/CBUA respondents provide data reference services to help users find and select research data.

● 62.5% of CAUL/CBUA respondents advise and/or provide instruction on how to cite data sources.

3CAUL/CBUA members requesting access to the detailed survey results should contact manager@caul-cbua.ca.

11 | Page

 

 

● 43.8% of CAUL/CBUA respondents promote a culture of data sharing and data reuse at their institutions through handouts, teaching, or participating in GIS Day or Open Access Week.

● 37.5% of CAUL/CBUA respondents offer services to reformat data for users to facilitate their use of data (e.g. converting files from SPSS to Microsoft Excel)

● 37.5% of CAUL/CBUA respondents offer services to transform data files for users (e.g. extracting data subsets, merging data files, or creating new variables)

● Two institutions, DAL and StFX, maintain a website that lists online research data management resources.

● DAL and UPEI maintain data curation profiles of their user communities. ● DAL and UPEI offer the following:

○ provide research data management training for faculty and/or graduate students; ○ recommend or provide instruction on the use of online tools for research data

management (e.g. Manta, DMPTool, etc.); and ○ assist researchers with preparing Data Management Plans.

Access Services Access services were described in the survey as “activities dealing with support needed to provide users with access to data collections and resources, including data platforms, data linkage, data retrieval, and data tools”. Some key observations are as follows:

● Over half (56.25%) of CAUL/CBUA respondents provide access to metadata discovery tools beyond their OPAC (e.g. Nesstar, DataVerse, or MarkLogic servers).

● 43.75% provide access to online data access tools such as an FTP or a DataVerse server. ● 37.50% provide access to software for analyzing and visualizing research data. ● 37.50% of CAUL/CBUA respondents support a local website that describes data and

contains links for downloading data. ● 31.25% provide access to online subsetting tools (e.g. Nesstar or SDA server). ● 31.25% of CAUL/CBUA respondents support a secure data enclave to provide research

access to sensitive data. ● UPEI and MUN provide access and/or support to data cleaning, processing, or format

translation tools (e.g. DataWrangler, Stat Transfer, or Google Refine). ● DAL is the only institution that links out to DataCite Canada. ● DAL is also the only institution that subscribes to the Data Citation Index through the Web

of Knowledge platform. ● There are no CAUL/CBUA institutions currently connecting local research data files with

their OPAC.

12 | Page

 

 

Preservation Services

Preservation services were described in the survey as “activities describing services to support the mid-term and long-term preservation of research data”. Key observations for preservation services are as follows:

● Four institutions (DAL, King’s College, MSVU, and UPEI) offer advice and help for researchers to locate an appropriate repository for their research data.

● DAL, UPEI, and MSVU assist researchers with the selection of appropriate data and metadata standards for data preservation.

● UPEI and Kings College are the only institutions to provide researchers with tools to submit their data and metadata for long-term preservation.

● UPEI and DAL are the only institutions to support and maintain preservation storage and management systems for the long-term preservation of data.

● UPEI is the sole institution to offer: ○ a research records retention policy that addresses the preservation and protection

of research data assets; ○ support for a staging repository for researchers to deposit data for short-term

storage and subsequent long-term deposit; ○ support and preparation of research data archival information packages for

long-term preservation. ● None of the institutions polled maintain a registry of acceptable or recommended file

types for research data and metadata. ● Additionally, none of the institutions polled have a formal data deposit agreement form

for researchers to sign when they submit their data.

13 | Page

 

 

SURVEY

Gaps and Requirements It is clear from the results that many CAUL/CBUA institutions have a lot of room for improvement in terms of research data management. The majority of institutions do provide access to research data through subscriptions. Even DAL and UPEI – with 90% and 73% affirmative responses, respectively – have room to grow. The most obvious gaps are specifically with technical services. Preservation, data management plans, and secure places to store content all appear as clear gaps within the region. The results show that many of the responding institutions offer something in the way of subscription services, reference help, and literature or guides of some sort to aid researchers; however, the survey does little to address the fact that some of the CAUL/CBUA institutions may actually see little utility in providing research data services of any kind. Indeed, in their comments section, NSCAD noted:

“The nature of research data in visual arts is an area still being figured out: As of yet, the NSCAD Library does not have any services to support research data collection but it is a subject I will discuss with our institution’s Library Committee.”

Similarly, Université Sainte-Anne noted that they would contribute where resources allowed, but that their size prevented them from taking much of a step forward at this juncture.

14 | Page

 

 

SURVEY

Summary and Discussion Research data stewardship is a fairly new field. Many schools in CAUL/CBUA are just beginning to gain momentum with general scholarly communications initiatives like Institutional Repositories. It is not surprising, then, that many institutions do not yet have well-developed research data services. The adapted CARL survey reveals that the region has a lot of room to grow. With the arrival of the Tri-Council Open Access policies it is a particularly relevant time to investigate 4

how CAUL/CBUA members can best respond to this national mandate. A collaborative and regional response that intersects with national activities, as well as one which promotes and supports local efforts are recommended by this committee. This survey strongly suggests that DAL and UPEI have important roles to play as regional leaders in research data stewardship. Their experience in research data management and preservation should, at least, provide some guidance for those institutions beginning to look at these issues. It would be prudent, however, to note that the survey itself may not offer the best measures of awareness or action on the topic of research data stewardship. For example, many of the questions from both the “access” and “user” sections of the survey relate to providing links to existing data repositories, databases, and other such services. It may be debated that linking to a resource is the same as providing access. Dataverse, for example, might not warrant a link on a library website if a liaison or scholarly communications librarian gets this information to researchers through other methods. We recognize that the survey may have missed additional ways that institutions are approaching issues surrounding research data stewardship. It is worth noting that results for Dalhousie are more current – by as much as two years – than other institutions. Their responses were updated in February 2015. In order to best reflect the ongoing efforts of the CAUL/CBUA members it would be beneficial to find a way to maintain a dynamic reflection of members services in this area.

4 Government of Canada (2015, February 27). Tri-Agency Open Access Policy on Publications. Retrieved from

http://www.science.gc.ca/default.asp?lang=En&n=F6765465-1

15 | Page

 

 

RECOMMENDATIONS AND NEXT STEPS Introduction

This section provides recommendations for the establishment and expansion of research data management services within the CAUL/CBUA consortium. The proposed expansion involves the establishment of an Atlantic Research Data Repository (ARDR), which would be jointly hosted by four CAUL/CBUA institutions, one in each province. The proposed repository would be made available to all CAUL/CBUA members. Specific recommendations regarding infrastructure and funding are outlined below. This section also contains recommendations regarding the distribution of the Research Data Stewardship Survey data, and the continuation of the survey.

16 | Page

 

 

RECOMMENDATIONS AND NEXT STEPS Survey Recommendations

The survey was intended to provide a current picture of research data stewardship services offered by each CAUL/CBUA institution. The survey results can provide valuable information to each institution, aiding in goal setting and providing a sense of the regional capabilities for data stewardship. For this reason, we recommend that the survey results be posted in full on the CAUL/CBUA website. As data stewardship becomes increasingly necessary, it will be useful to track our regional and institutional capabilities in the coming years. Thus, we recommend that the survey be completed on an annual or biennial basis. The survey in its current form is not ideal, so we recommend that the survey questions be evaluated and updated as necessary for each survey cycle. While this may make some data points difficult to track year over year, it will allow us to eliminate or modify questions that become irrelevant as regional and technological capabilities grow. For the next cycle of the survey, we recommend employing a survey tool (such as Survey Monkey) which can provide a long term framework for maintaining the survey. A survey tool will allow for a better respondent experience, and facilitate results analysis. We also recommend offering a response category labeled “in progress” to allow institutions to indicate services that are in development, but not yet fully realized. Lastly, it would be useful to provide more opportunities for comments within the survey, to allow respondents to clarify or elaborate on answers to specific questions.

17 | Page

 

 

RECOMMENDATIONS AND NEXT STEPS

Research Data Management Infrastructure

CAUL/CBUA libraries have a great opportunity to assist researchers with data management; however, this assistance will require a robust infrastructure. The following sections will outline the necessary components of such an infrastructure including: a RDMP tool, data repository services (data storage, metadata creation, and search interface), the creation of a CAUL/CBUA RDM team, appropriate governance, and recommendations for a funding model. The development of this project should dovetail with similar endeavors like Portage, benefiting from national infrastructure efforts.

Research Data Management Planning Tool

The TC3+ will soon mandate that all grant applicants provide a research data management plan (RDMP). We feel that our libraries should offer support in the creation of these plans. There are tools in development at a number of institutions in North America for creating RDMPs. We recommend that CAUL/CBUA endorse one of these tools, with an eye to the RDMP tool currently under development at the University of Alberta, CARL and Compute Canada. Individual CAUL/CBUA institutions may also wish to deploy local instances of a similar tool in order to better reflect institutional practices. In these cases, we recommend that locally deployed tools be able to export plans to the national repository, contributing to a national inventory of RDMPs. Providing a RDMP tool to our researchers will require resources (funding, staffing, etc.). As it develops, the Portage model may help to inform some of these resource decisions and may even provide some funding for the ARDR initiative, depending on how CAUL/CBUA decides to participate in this project. If the CAUL/CBUA proceeds with a project similar to ARDR, then CAUL/CBUA libraries and staff would develop strong regional expertise, benefiting not only the region, but national efforts as well.

Regional Research Data Storage Service

Whether or not CAUL/CBUA decides to provide a regional data storage solution will depend upon a desire to actively and collectively steer data management practices. National projects such as Portage are on the horizon, and some individual CAUL/CBUA institutions are providing strong RDM services. The establishment of an Atlantic Research Data Repository would allow regional capabilities to grow more quickly, and would help distribute the financial burden of these services across all member institutions. Data storage is a key requirement of a RDM service, whether providing secure working storage or long-term preservation storage. Individual research projects will vary widely as to the specific requirements for the amount and type of storage, including the potential for encrypted storage

18 | Page

 

 

for data containing private information, Terabyte/Petabyte-level massive storage, efficient data analysis tools, and more. Requirements common to all research projects include: secure and redundant storage which can ensure data integrity and preservation for the short and long term; descriptive and administrative metadata; data publishing services. One cost-effective approach to a sustainable data storage service would be for CAUL/CBUA institutions to collaborate on the provision and delivery of storage services. We recommend CAUL/CBUA support the creation of the ARDR service, which would function on an “all-in” cost-sharing model and would provide data storage and services to all member institutions. The ARDR system would provide a two-tiered service model, offering a basic service package as well as a value-added service package. Member institutions would be free to choose the package that best suited their needs. The basic service package would include:

● creation and storage of a final dataset, defined by the Principal Investigator as one needing an appropriate level of accessibility (e.g. is part of a publication, or needs to be shared with collaborators);

● creation and vetting of a Dublin Core record for the dataset; ● creation of a VIVO/CASRAI (or similar) record to identify the

institutional/researcher/funder context; ● minting of a DOI for the dataset; ● synchronization/integration of the dataset with national and/or domain-specific

repositories; ● support for library staff and researchers in all aspects of research data management.

The value-added service would include:

● bulk storage of active data; ● access to a DropBox-style service (but one with provincially provisioned storage) that

allows a researcher to easily synchronize active data from a local desktop or system to a central managed service;

● other services deemed useful to researchers and consistent with member institution’s practices.

The value-added service would come at an additional cost to institutions that choose this option. In regards to additional data storage as an example, costs would vary depending on the amount of storage required. The RDWG proposes that the hardware infrastructure (i.e. servers, storage drives, etc.) be hosted in four locations: UPEI, UNB, Dalhousie, and MUN. These four sites would serve as regional nodes, including providing back-ups for one another, with all data synchronized between sites for added redundancy. Given the costs of setting up and maintaining this hardware,

19 | Page

 

 

our proposed funding model would funnel a significant portion of the member fees to the four host institutions.

CAUL/CBUA RDM Team

A vital component of the ARDR proposal is the development and sharing of expertise in the area of research data management. CAUL/CBUA members are already cultivating this expertise in their own institutions, but experts within separate institutions can benefit from formalized information sharing within CAUL/CBUA. We recommend the establishment of a CAUL/CBUA Research Data Management Team. This team would consist primarily of the CAUL/CBUA Manager and one or more additional CAUL/CBUA employees with technical expertise in data management. Funding for the additional position(s) could come from CAUL/CBUA membership fees, individual research projects, or external subsidies from larger research data management programs like Portage. This team would offer RDM support to member institutions. The RDM Team would provide annual training workshops in matters relating to RDM, as well as offer ongoing support via a mechanism such as instant messaging or a ticketing system similar to those commonly used in IT support. This team would be accessible to faculty and staff at all member institutions. The RDM Team would report to the CAUL/CBUA Directors via the DPSC. In addition to the CAUL/CBUA RDM Team, we recommend establishing a support network akin to the current Data Liberation Initiative (DLI) model. This would see each member institution designating a RDM representative who would liaise with the RDM Team as well as representatives from other member institutions. The RDM Team would provide a mailing list where RDM representatives could share information, post questions, and offer advice to one another.

ARDR Governance and Administration

The ARDR will require a strong RDM policy framework to guide implementation and development. This framework will have to account for institutional autonomy, but CAUL/CBUA could offer model policies for institutions to consider as they develop their internal services. All of the following policy proposals are aspirational, and full implementation will take several years. In the development of model policy, CAUL/CBUA should examine and adapt policies at place in other research networks. Specifically, we recommend adopting a model policy that CAUL/CBUA would endorse for the oversight of ARDR, and individual institutions would use as the basis for local policies. The example used here is based on the the University of Edinburgh’s RDM policy . 5

5 University of Edinburgh, The. (2015, February 5). Research data management policy. Retrieved from http://www.ed.ac.uk/schools-departments/information-services/about/policies-and-regulations/research-data-policy

20 | Page

 

 

The aforementioned policy contains the following ten guiding points, slightly modified here to better suit CAUL/CBUA:

1. Research data will be managed to the highest standards throughout the data lifecycle as part of the institution’s commitment to research excellence.

2. Responsibility for research data management through a sound research data management plan during any research project or programme lies primarily with Principal Investigators (PIs).

3. All new research proposals [from date of adoption] must include research data management plans or protocols that explicitly address data capture, management, integrity, confidentiality, retention, sharing, and publication.

4. CAUL/CBUA will arrange training, support, advice, and where appropriate guidelines and templates for the research data management and research data management plans.

5. CAUL/CBUA will facilitate the provision of mechanisms and services for storage, backup, registration, deposit and retention of research data assets in support of current and future access, during and after completion of research projects.

6. Any data which are retained elsewhere, for example in an international data service or domain repository, should be registered with the institution.

7. Research data management plans must ensure that research data are available for access and re-use where appropriate and under appropriate safeguards. If possible, data should be made accessible with a statement like this:

a. To the extent possible under law, the authors have waived all copyright and related or neighbouring rights to this data. CC0/Open Data.

8. The legitimate interests of the subjects of research data must be protected. 9. Research data of future historical interest, and all research data that represent records of

a member institution, including data that substantiate research findings, will be offered and assessed for deposit and retention in an appropriate national or international data service or domain repository, or a member institution’s repository.

10. Exclusive rights to reuse or publish research data should not be handed over to commercial publishers or agents without retaining the rights to make the data openly available for re-use, unless this is a condition of funding.

Another important policy statement, not included in the model policy above, pertains to the duration of data retention. We recommend that CAUL/CBUA institutions adopt a minimal RDM preservation commitment that says: “The University will steward the data for [Project X] for as long as is needed”. This statement will allow for flexible retention timelines that can be adjusted in consultation with the PI. Given the active discourse around RDM in Canada and further afield, we recommend that all ARDR services and policies be crafted with an awareness of national RDM initiatives. A well-designed ARDR should be poised to integrate with CARL’s Portage and the Compute Canada cyberinfrastructure.

21 | Page

 

 

ARDR Sustainability Should the CAUL/CBUA Directors decide to pursue the ARDR project, a sustainable funding model will have to be established. We are proposing an initial 3-year financial investment from all CAUL/CBUA members to get the project started. After the initial 3-year period, research grant money can be leveraged to provide ongoing funding. According to the Association of Atlantic Universities (AAU), Atlantic universities generate $500 million in research funding annually, approximately 75% of which is TC3+ funding. If we were to assume a TC3+ mandate for research data stewardship, even 1% of the TC3+ grant money ($3,750,000) could keep the ARDR sustainably funded, negating the need for further financial contributions from CAUL/CBUA members. Despite this potential funding base, we recommend that CAUL/CBUA look for grant opportunities (ACOA, TC3+, etc.) to assist with the start-up costs. While not prescriptive, a basic funding proposal is included below to suggest one possible approach to funding ARDR in the near term.

1) Cost Model a) CAUL/CBUA RDM staff person

i) Annual requirements: FT equivalent resource ($90,000), travel funds ($10,000) ii) One-time costs: Laptop and software ($5,000)

b) Infrastructure i) Support for centralized ARDR repository: $60,000

c) Other: $40,000 d) Total Annual: $200,000

2) Funding Model

a) Core Services (multiple options) i) Using current CAUL/CBUA membership model, member fees for the initial 3

years would be as follows:

INSTITUTION FEE

Acadia University $10,600

Atlantic School of Theology $200

Cape Breton University $7,200

Dalhousie university $44,100

Holland College $5,200

Memorial University of Newfoundland $42,120

Mount Allison University $6,380

Mount Saint Vincent University $7,160

NSCAD University $1,860

NSCC $6,380

Saint Mary’s University $17,120

22 | Page

 

 

St. Francis Xavier University $11,400

Université de Moncton $13,160

Université Sainte-Anne $1,000

University of King’s College $3,000

University of New Brunswick Fredericton $18,860

University of New Brunswick Saint John $5,580

University of Prince Edward Island $10,160

ii) Tiered membership model (variable cost) iii) Equal costs per institution all-in ($11,000 per year for 3 years) iv) Equal costs per institution opt-in (variable cost)

b) Revenue Model

This model assumes that those institutions providing the shared services (hosting hardware/software infrastructure and providing the staff to maintain said infrastructure) would split the ARDR infrastructure funds to help offset the costs of local services. Using a 4-institutions model in this example, the revenue sharing would provide $15,000 annually to the 4 host institutions.

A firm commitment to this ARDR vision from CAUL/CBUA members will allow CAUL/CBUA to play a significant and active role in the guidance and development of research data management in the region and at a national level. A robust ARDR could become an important component of Portage and provide a Canadian model for regional RDM. The proposed 3-year investment will help member institutions develop the necessary expertise to stand at the forefront of research data management and allow Atlantic Canadian researchers to continue to excel at national and international levels.

23 | Page

 

 

APPENDIX A: Glossary and Acronyms

Access Services: Activities dealing with support needed to provide users with access to data collections and resources, including data platforms, data linkage, data retrieval, and data tools. Aggregate data: The organization of statistics into a data structure, to store in a database or in a data file. ARC: Advanced Research Computing. A project initiated by the Canadian Association of Research Libraries (CARL), aiming to continue work on research data management in Canada. Born Analog: Information that was originally created in a non-digital format and has been digitized. Born Digital: A digital object that has never had an analog form. CADC: Canadian Astronomy Data Centre CARL: Canadian Association of Research Libraries CASRAI: Consortia Advancing Standards in Research Administration Information CBRAIN: Web-based software that allows neuroimaging researchers to perform analyses on data by connecting to High-Performance Computing facilities. CFI: Canada Foundation for Innovation CIHR: Canadian Institutes of Health Research Collection Services: Activities that specifically support the development, acquisition, management, description, and discovery of a collection of research data files. Compute Canada: Compute Canada deploys state-of-the-art advanced research computing (ARC) systems, storage and software in partnership with regional organizations ACENET, Calcul Quebec, Compute Ontario and WestGrid. CRKN: Canadian Research Knowledge Network CUCCIO: Canadian University Council of Chief Information Officers Dark Archive: An archive that does not grant public access. Data: Facts, ideas, or discrete pieces of information, especially when in the form originally collected and unanalyzed. Data Management Plan: A data management plan is a formal document that outlines what you will do with your data during and after a research project. 6

DC/QDC: Dublin Core / Qualified Dublin Core. A standard for metadata description. Digital object: A representation of information in digital form. Digital Preservation: The series of management policies and activities necessary to ensure the enduring usability, authenticity, discoverability and accessibility of content over the very long term. The key goals of digital preservation include usability, authenticity, discoverability, and accessibility. 7

6 DMP Tool, Data Management General Guidance. Web. Accessed March 2015. https://dmptool.org/dm_guidance 7 Portico. Web. Accessed March 2015. http://www.portico.org/digital-preservation/glossary 24 | Page

 

 

DILC: Digital Infrastructure Leadership Council. DILC acts as a forum to discuss, develop, and coordinate Canada’s digital infrastructure networks. DLI: Data Liberation initiative. DMP: Data Management Plan. DMTI: DMTI Spatial. A provider of digital mapping data, location based data, geocoding, routing and GIS software. DPSC: Digital Preservation and Stewardship Committee. A committee of the Council of Atlantic University Libraries. DRM: Digital Rights Management GIS: Geographic Information Systems (or Science). Software that allows the visualization and analysis of spatially referenced data on spatial data. GIS Day: This is a grassroots educational event promoting the use of GIS and showcasing the uses of GIS. ICPSR: Interuniversity Consortium for Political and Social Research IDSE: Integrated Digital Scholarship Ecosystem. A CRKN initiative aiming to provide guidelines for digital scholarship in a Canadian research context. IPY: International Polar Year Project ISO: International Organization for Standardization LCDI: Leadership Council for Digital Infrastructure Member Institutions: Also referred to as Members. The post-secondary libraries belonging to the Council of Atlantic University Libraries. Metadata: data that describes information about digital objects. Descriptive/Bibliographic Metadata: Information used to search and locate an object such as title, author, subjects, keywords, and publisher Technical Metadata: Information about aspects of the object related to its file format or the original software used to create the file. Administrative Metadata: Information needed to help manage the digital object, such as copyright and preservation information. Structural Metadata: Information on how the digital object is organized, including the pages, chapters, and indexes. Methodology: The procedure(s) used to collect information or research, which explains the scope of the study, including factors such as sample selection, data sources, and disclosure. METS: Metadata Encoding and Transmission Standard. A framework for describing metadata. Migration: Process of changing a file format. NSERC: Natural Sciences and Engineering Research Council OAIS: Open Archival Information System. Archival framework developed by the Consultative Committee for Space Data Systems (CCSDS). Obsolete Format/Technology: Hardware or software that is no longer widely used. OPAC: Online Public Access Catalogue

25 | Page

 

 

Open Archival Information System (OAIS): An archive that meets a set of responsibilities, as defined in the OAIS Reference Model. Preservation Description Information (PDI): The information necessary for preservation purposes, such as provenance, reference, context, and access rights information. Preservation Services: Activities describing services to support the mid-term and long-term preservation of research data. Provenance Information: Information that documents the history of an object, including its origin, changes that may have occurred over the course of its life cycle, and current custody. Digital Provenance is information regarding the origin of a digital object. RDC: Research Data Canada RDM: Research Data Management RDMPT: Research Data Management Pricing Tool RDWG: Research Data Working Group. A Working Group under the CAUL/CBUA DPSC. Research Data Stewardship: The management and care of research data. Render: To process a digital object, in order to view, listen to, or interact with the content. Repository: An area designated to storing and maintaining items. Digital repositories house digital objects. Research Data Management Planning: A plan outlining the storage, maintenance, management, and policies relating to research data. SOA: Service Oriented Architecture SPSS: A statistical analysis package with data management functions. SSHRC: Social Sciences and Humanities Research Council Succession Plan: A procedure outlining how and when to transfer the management, ownership and/or control of holdings. TC3+: Tri-Agency Council. Comprised of the Social Sciences and Humanities Research Council (SSHRC), the Natural Sciences and Engineering Research Council (NSERC), the Canadian Institutes of Health Research (CIHR), in collaboration with the Canada Foundation for Innovation (CFI) and with Genome Canada. TDR: Trustworthy Digital Repository TRAC: Trustworthy Repositories Audit and Certification User Services: Activities that focus on supporting user communities by identifying their data needs, assisting them in preparing data management plans, selecting metadata standards and best practices, identifying existing data sources, and retrieving, manipulating, and transforming data.

26 | Page

 

 

VIVO: An open-source software system, a network of investigators and institutions, and an open information representation model for scholarship. VIVO leverages work done over the past nine years by Cornell University, supporting researchers and finding of researchers by representing data about them and their activities including publications, awards, presentations and partners. Support for researchers using VIVO is often done by librarians of the research institutions. 8

Workflow: The formalization of the process metadata which includes a description of the researcher's method. It identifies the data inputs, transformations, and analytical steps to achieve the final data output. 9

8 University of Nebraska Medical Center. Data Management. Web. Accessed March 2015. http://unmc.libguides.com/content.php?pid=525776&sid=4325759 9 US Geological Survey. (2015, January 16). USGS Data Management. Retrieved from http://www.usgs.gov/datamanagement/describe/capture.php 

27 | Page

 

 

APPENDIX B: Bibliography

Baker, K. S., & Yarmey, L. (2009, October 15). Data stewardship: Environmental data

curation and a web-of-repositories. International Journal of Digital Curation, 4, 2, 12-27. DOI:10.2218/ijdc.v4i2.90

Berman, F. (2008, December 01). Got data?: A guide to data preservation in the information

age. Communications of the ACM, 51, 12, 50-56. DOI: 10.1145/1409360.1409376 CARL. (2014, December 22). Portage: Supporting Canadian innovation through shared expertise and stewardship of research data. Retrieved from http://www.carl-abrc.ca/uploads/SCC/Portage-External-2-Dec-22-2014.pdf CARL. (2013, December 12). CARL’s response to the consultation document Capitalizing on Big Data:

Toward a Policy Framework for Advancing Digital Scholarship in Canada. Retrieved from http://www.carl-abrc.ca/uploads/SCC/CARL%20Big%20Data%20Consultation%20Response%20Dec%2012%202013.pdf

Delserone, L. M. (2009, March 27). At the watershed: Preparing for research data management and stewardship at the University of Minnesota Libraries. Library Trends, 57, 2, 202-210. DOI: 10.1353/lib.0.0032

Government of Canada. (2015, February 27). Tri-Agency Open Access Policy on Publications. Retrieved from

http://www.science.gc.ca/default.asp?lang=En&n=F6765465-1 Government of Canada. (2014). Canada’s Action Plan on Open Government 2014-16. Retrieved from http://open.canada.ca/en/content/canadas-action-plan-open-government-2014-16 Hedstrom, M. (1998, January 01). Digital preservation: A time bomb for digital libraries.

Computers and the Humanities, 31, 3, 189-202. Retrieved from http://www.uky.edu/~kiernan/DL/hedstrom.html

Heery, R. and Anderson, S. (2005) Digital repositories review. University of Bath. Retrieved from

http://opus.bath.ac.uk/23566/2/digital-repositories-review-2005.pdf

Research Data Strategy Working Group. (2008). Stewardship of research data in Canada: A gap analysis. Retrieved from http://rds-sdr.cisti-icist.nrc-cnrc.gc.ca/eng/reports/2008_gap_analysis.html

Rosenbaum, S. (2010, October). Data governance and stewardship: Designing data

stewardship entities and advancing data access. Health Services Research, 45, 5, 1442-55. DOI: 10.1111/j.1475-6773.2010.01140.x

Shearer, K., & Canadian Association of Research Libraries. (2009). Research data: Unseen

opportunities. Ottawa, Ont: Canadian Association of Research Libraries. Retrieved from http://carl-abrc.ca/uploads/pdfs/data_mgt_toolkit.pdf

28 | Page

 

 

SSHRC, NSERC, CIHR, and CFI. (2013, October 16). Capitalizing on big data: Toward a policy framework

for advancing digital scholarship in Canada. Retrieved from http://www.sshrc-crsh.gc.ca/about-au_sujet/publications/digital_scholarship_consultation_e.pdf

University of Edinburgh, The. (2015, February 5). Research data management policy. Retrieved from http://www.ed.ac.uk/schools-departments/information-services/about/policies-and-regulations   /research-data-policy

29 | Page

 

 

APPENDIX C: CAUL/CBUA Research Data Management Survey Instrument

Objective The purpose of this exercise is to benchmark your Library’s current involvement in data management services. This is not a comprehensive list of activities that a Library could undertake in providing data management services, but rather is a sample of items that we feel is representative of such activities. This survey is shamelessly copied from the CARL instrument with permission. If your institution responded to this CARL survey, please feel free to forward it rather than filling this out again, as they essentially the same document.

Please read each of the ten items under the four service areas and Highlight the items that your Library currently carries out or supports. When completed, add the number of highlighted items within each service area and then determine the total for all service areas. Use the Score Sheet below to record these sums. Please also feel free to add any comments you have to each item where appropriate. We are especially interested in details that can help us start an inventory of what services and resources CAUL members provide for research data management.

The DPSC Research Data Working Group will discuss the results at their meeting at APLA in June. In order to facilitate that review we would ask that you respond by filling out this document by May 16. If you have any questions please contact Mark Leggott - 902-566-0460, mleggott@upei.ca.

A glossary is provided to clarify the meaning of terms used.

Collection Services These are activities that specifically support the development, acquisition, management, description, and discovery of a collection of research data files.

1. Do you have a collection policy for research data? 2. Do you have subscriptions with data providers (e.g. DLI, DMTI, or ICPSR)? 3. Do you have a dedicated budget to purchase data files outside of subscription services? 4. Do you maintain a collection of data files from local researchers? 5. Do you catalogue local research data files in your OPAC? 6. Do you maintain infrastructure to manage a local data file collection (e.g. a digital assets

management system)? 7. Do you produce data documentation to enhance your data collection (e.g. user’s guides,

data dictionaries, variable lists, etc.)? 8. Do you have metadata librarians or specialists who advise on standards for content and

technical metadata? 9. Do you produce standards-based metadata for research data?

10. Is your department or unit a member of a standards body for research data or metadata?

User Services These are activities that focus on supporting user communities by identifying their data needs, assisting them in preparing data management plans, selecting metadata standards and best practices, identifying existing data sources, and retrieving, manipulating, and transforming data.

30 | Page

 

 

1. Do you collect and maintain data curation profiles of your user communities? 2. Do you conduct activities that promote a culture of data sharing and data reuse at your

institution, e.g., through handouts, teaching, or participating in GIS Day or Open Access Week?

3. Do you provide research data management training for faculty or graduate students? 4. Do you recommend or provide instruction on the use of online tools for research data

management (e.g. Mantra, DMPTool, etc.)? 5. Do you assist researchers with preparing Data Management Plans? 6. Do you maintain a website that lists online research data management resources? 7. Do you advise on or provide instruction on how to cite data sources? 8. Do you provide data reference services to help users find and select research data? 9. Do you reformat data for users to facilitate their use of data (e.g. convert from SPSS to

Excel)? 10. Do you transform data files for users (e.g. extract data subsets, merge data files, or create

new variables)?

Access Services This set of activities deals with the support needed to provide users with access to data collections and resources, including data platforms, data linkage, data retrieval, and data tools.

1. Do your OPAC records provide links to local research data files? 2. Do you support a local website that describes data and contains links for downloading

data? 3. Do you provide a link to DataCite Canada from your local website? 4. Do you subscribe to the Data Citation Index through the Web of Knowledge platform? 5. Do you provide metadata discovery tools beyond an OPAC (e.g. Nesstar, DataVerse, or

MarkLogic server)? 6. Do you provide online data access tools (e.g. FTP or DataVerse server)? 7. Do you provide access to online data subsetting tools (e.g. Nesstar or SDA server)? 8. Do you provide access and support to data cleaning, processing, or format translation

tools (e.g. DataWrangler, Stat Transfer, or Google Refine)? 9. Do you provide access to software for analyzing and visualizing research data?

10. Do you support a secure data enclave to provide research access to sensitive data?

Preservation Services These activities describe services to support the mid-term and long-term preservation of research data.

1. Does the University have a research records retention policy that addresses the preservation and protection of research data assets?

2. Does the library have a mandate to preserve research data? 3. Do you advise on or help researchers locate an appropriate repository for their

research data? 4. Does the library have a formal data deposit agreement form for researchers to sign

when they submit their data? 5. Do you assist researchers with the selection of appropriate data and metadata

standards for the preservation of data? 6. Do you maintain a registry of acceptable or recommended file types for research data

31 | Page

 

 

and metadata? 7. Does your library support a staging repository for researchers to deposit data for

short-term keeping and subsequent long-term deposit? 8. Does your library provide researchers with tools to submit their data and metadata

for long-term preservation? 9. Does your library support and prepare research data archival information packages

for long-term preservation? 10. Does your library support and maintain preservation storage and management

systems for the long-term preservation of data?

Score Sheet

Function Number of Activities Circled

Collection Services

User Services

Access Services

Preservation Services

TOTAL

General Comments Please record any general comments you wish us to consider here.

Survey Glossary Archival Information Packages: A concept from the OAIS Reference Model describing the package of digital objects organized, documented, and managed in a long-term preservation environment. Data Citation Index: A Thompson Reuters database linking data files in repositories with published literature that cites data. DataCite Canada: An online data registry service provided by the National Science Library of the National Research Council to assign digital object identifiers (DOIs) to data files. i.e. persistent and unique identifiers for data. Data Curation Profiles: Narrative-based methodology for describing research data from individual or team research projects. The Purdue Data Curation Profile is one approach to documenting researchers' data management activities, their data holdings and their data management practices. The Digital Curation Centre Data Asset Framework is another method. Data Deposit Agreement: A document specifying the terms and responsibilities of the researcher depositing her or his data with a repository and the terms and responsibilities of the repository in disseminating the data. Data Dictionary: Supporting data documentation for a data file that identifies variable names and labels, origin of the variable, values and labeling, missing data codes, record layout, and other related information. Data Enclave: A secure facility for analyzing sensitive data. Services commonly associated with a data enclave include restricted and authenticated access to the facility and disclosure approval for analysis results to be removed from the facility.

32 | Page

 

 

Data Management Plans: A formal document outlining how a researcher plans to handle her or his data both during research and after a project is completed. Data Subsets: Customized extractions from a complete data file consisting of selected cases (observations) or variables. DataVerse: A virtual data collection system for managing and retrieving data files. DataWrangler: A software product for interactively cleaning and transforming data. DLI: Data Liberation Initiative, a subscription program between Statistics Canada and post-secondary institutions providing access to all standard data products and spatial data. DMPTool: An online tool developed by the California Digital Library to produce data management plans. DMTI: A spatial data collection. Excel: Microsoft’s spreadsheet program distributed as part of the Microsoft Office Suite. FTP: A file transfer service based on the file transfer protocol. SFTP (secure file transfer protocol) has tended to replace earlier FTP services. GIS: Geographic Information Systems (or Science). Software that allows the visualization and analysis of spatially referenced data on spatial data. GIS Day: This is a grassroots educational event promoting the use of GIS and showcasing the uses of GIS. Google Refine: A Google cloud-based tool for editing messy data, transforming it to other formats, and providing access the data through web services. ICPSR: Inter-university Consortium for Political and Social Research, a large membership-based data repository for social science data. Mantra: An online instructional course developed and maintained by EDINA at the University of Edinburgh based on best practices in research data management from three disciplines: social science, clinical psychology, and geoscience. OPAC: Online Public Access Catalogue Open Access Week: This is a grassroots educational event promoting Open Access publishing. MarkLogic: A commercial database system capable of indexing structured, semi-structured, and unstructured digital content. Metadata: Descriptive information about other digital objects. Some metadata are based on a standard, while other metadata are based on local convention. Nesstar: A Web-based service for data discovery and dissemination. Sensitive Data: A data file containing information that could easily disclose the identity or location of an observation within a data file. For example, the names, street addresses, or phone numbers of individuals in a file make the data sensitive. The location of nesting grounds for an endangered species in a data file is also sensitive data. SDA: A set of Web-based programs for the documentation and analysis of survey data. SPSS: A statistical analysis package with data management functions. Staging Repository: A service for organizing and submitting research data for a period of time to provide immediate access to the data. A staging repository may work with a data repository supporting long-term preservation services and have arrangements to structure the data it holds for submission with the long-term preservation repository. Stat Transfer: A program for changing the formats between popular software systems. User Communities: The groups of data users sharing a common background, such as, discipline, data source (e.g., all Census users), or authorization category (e.g., all graduate students.) User’s Guide: Supporting data documentation that provides a description of the study or program under which the data were produced, the study design, sampling methodology, data collection and editing process, weighting procedures, and other related information.

33 | Page

 

 

Appendix D: CAUL/CBUA Survey Infographic

34 | Page

 

CAUL/CBUA RESEARCH DATASTEWARDSHIP SERVICESSURVEY: RESULTSThe CAUL/CBUA Digital Preservation Stewardship Committee (DPSC) Research DataWorking Group (RDWG) was created in Winter 2014 for the purpose of conductingresearch related to Research Data Stewardship in the Atlantic Region. DPSC RDWGmembers determined that a survey would be the best tool for collecting reliable dataand staff input from across the CAUL/CBUA community. A survey distributed by CARLwas adapted with permission and distributed to University Directors on April 13, 2014.

16/17 InstitutionsResponded to the survey.

CAUL/CBUA Members

The SurveyThe survey itself focused primarily on four aspects of research data stewardshipservices, with 10 questions in each section: 

Collection Services“Activities that specifically support the development,acquisition, management, description, and discovery of acollection of research data files”.

User Services“Activities that focus on supporting user communities by

identifying their data needs, assisting them in preparingdata management plans, selecting metadata standards andbest practices, identifying existing data sources, andretrieving, manipulating, and transforming data”.

Access Services“Activities dealing with support needed to provide userswith access to data collections and resources, includingdata platforms, data linkage, data retrieval, and data tools”.

Preservation Services“Activities describing services to support the mid-term andlong-term preservation of research data”.

Results at a GlanceResults for types of services offered at CAUL/CBUA institutions were quite varied. Mostinstitutions were able to provide at least token Collections, User, or Access services.However, even the highest scoring institution in the survey – UPEI – answered in thenegative for 28% of the questions. Very few institutions are well served in terms ofPreservation services. Some institutions responded as having zero services in place atall, and only one failed to respond in general. 

The following graph is a representation of the total percentage of "yes" answers on thesurvey against the total number of questions for each institution that had at least oneaffirmative response.

Collections Services User Services Access Services Preservation Services

Total

ACA

MSVU

NSCC

DAL

SMU

STFX

UdeM

UNB

UNBSJ

UPEI

MUN

AST

UKC

% of affirmative answers for each section of the survey.0 10 20 30 40 50 60 70 80 90

Clearly, Preservation Services are far­and­away the most lacking amongst theCAUL/CBUA membership. It is also certainly worth noting that the institution with thehighest research data stewardship ratings also develops software that is particularlygood at meeting preservation needs. 

Every institution has room for improvement. 

Dal, generally, is the regional leader in research data stewardship efforts.

Some Additional Observations

3Institutions – CBU, USA, and NSCAD – reported no

activities related to research data stewardship.

"The nature of research data in visual arts isan area still being figured out".

- NSCAD Survey Comment

5 InstitutionsReported minimal activities related to research datastewardship. (ACA, AST, NSCC, StFX, UKC)

2 InstitutionsRecorded affirmative answers for > 50% of the survey.(DAl, UPEI)

3 InstitutionsHave mandates for the preservation of research data. (DAL,UPEI, MSVU)

Collection Services GapsWhile three of the schools – DAL, MUN, and UPEI – answered in the affirmative for70% of Collections questions, the rest of the region lagged behind significantly. Takingout the schools with no current activities in Research Data Stewardship, the averageresponse for Collections Services was only 1 affirmative answer out of 10. 

70% of InstitutionsHave subscriptions to data providers such as DLI, DMTI orICPSR.

UPEIUPEI is the only CAUL/CBUA institution with a writtenCollection Policy for research data.

DALDAL is the only CAUL/CBUA institution with membershipto a standards body for research data or metadata.

0%Of CAUL/CBUA member institutions are cataloguing localresearch data files in their OPAC.

User Services GapsUser services were fairly well represented across the CAUL/CBUA group, with only four

institutions falling short of providing any specific services. Most institutions are makingat least a gesture or two in this field, be it through Open Access Week events,reference help, or helping users to find research data. 

Most of the obvious gaps fall directly in the field of Data Management plans or specifictechnical assistance in modifying or converting specific types of data files. 

2 InstitutionsUPEI and Dal are the only institutions to provide researchdata management training for faculty and/or graduatestudents.

2 InstitutionsUPEI and Dal are also the only institutions recommendingor providing instruction on the use of online tools forresearch data management (e.g. Mantra, DMPTool, etc.).

2 InstitutionsUPEI and Dal are also the only institutions assistingresearchers with preparing Data Management Plans.

Access Services GapsGaps in access services are a little more spread out than user and Collections Services.It's notable that UPEI and DAL were much closer to other member institutions in thisparticular category. 

It's worth noting that most of the questions about Access Services regarded directlinking or access to a service that exists elsewhere. This doesn't mean that librarians orfaculty at these institutions cannot access this material and, really, only speaks tospace for linking on a website. 

7 InstitutionsHave no current Access Services as represented in thesurvey. That's nearly half of the respondents!

5 InstitutionsProvide both software for analysing and visualizingresearch data and also support a secure data enclave forstoring sensitive data.

Consolation PrizeAt least 9 institutions – more than half – either link out tosome existing resources or provide specific access to someresearch data discovery tools.

Preservation Services GapsPreservation is the space with the most room to grow, but is also likely the mostresource and infrastructure­dependent service. It is vital to note that research data

preservation is a fairly new problem for libraries. This may speak to the absence ofservices in this section.

Only 5 Institutions...... offer anything in the way of Preservation Services. Thisincludes preservation policies, helping researchers findappropriate repositories, application of standards tometadata, tools for submitting data, archival packages, andpreservation storage/management systems.

That means...... there are 12 CAUL/CBUA institutions with no researchdata Preservation Services.

UPEI AND DALUPEI and Dal are the only institutions offering extensivePreservation Services.

SummaryResearch data stewardship is a fairly new field. Many schools in the CAUL/CBUA groupare only just starting to get momentum with general scholarly communicationsinitiatives like Institutional Repositories, so it is not overly surprising that many do notyet have well­developed research data services. That said, the adapted CARL surveysuggests that the region has a lot of room to grow. With the eventual arrival of Tri­Council Open Access policies that lean increasingly towards access to research data, itis a particularly relevant time to be investigating how CAUL/CBUA members can bestrespond with either their immediate resources or a collaborative, regional effort. 

This survey strongly suggests that DAL and UPEI have important roles to play asregional leaders in research data stewardship. Their experience in this work should, atleast, provide some guidance for those institutions only just beginning to look at theissues. 

 

APPENDIX E: Comprehensive Brief on Research Data Management Policies (Kathleen Shearer)

Note: This is a final draft document provided by Kathleen Shearer for CAUL Directors and is not to

be redistributed. The final approved version will be released by the Tri-Councils in the near future.

35 | Page

 

 

Comprehensive Brief on Research Data Management

Policies

April 7, 2015

Prepared by Kathleen Shearer, Consultant

2

Table of Contents List of Acronyms……………………………………………………………… Pg. 3 1. Executive Summary…………………………………………………………. Pg. 4 2. Introduction…………………………………………………………………… Pg. 6 3. Policy Environment………………………………………………………….. Pg. 7 3.1 Policy Objectives and Principles…………………………………………. Pg. 7 3.2 Typical Policy Requirements……………………………………………… Pg. 8 3.3 International………………………………………………………………… Pg. 9 3.4 Canada……………………………………………………………………… Pg. 13 3.5 Other Stakeholders………………………………………………………… Pg. 17 4. Data Management Plans…………………………………………………… Pg. 19 4.1 Examples of DMP Requirements………………………………………… Pg. 20 5. Administering Policies………………………………………………………. Pg. 22 5.1 Policy Guidance………………………………………………………........ Pg. 22 5.2 Evaluating DMPs…………………………………………………………… Pg. 23 5.3 Monitoring Compliance………………………………………………........ Pg. 25 5.4 Confidentiality and Intellectual Property…………………………………. Pg. 26 5.5 Multiple Policies………………………………………………………......... Pg. 27 5.6 Costs………………………………………………………………………… Pg. 28 6. Approaches to Policy Implementation…………………………………….. Pg. 29 6.1 Engineering and Physical Sciences Research Council………………... Pg. 29 6.2 Pilot on Open Research Data (European Commission)……………….. Pg. 30 6.3 Code of Conduct and the Australian National Data Service…………... Pg. 31 6.4 Office of Science and Technology Policy (US)……………………........ Pg. 31 7. Implementation Challenges………………………………………………… Pg. 32 7.2 Disciplinary Context……………………………………………………….. Pg. 32 7.3 Research Preparedness…………………………………………………... Pg. 32 7.4 Incentives………………………………………………………………….... Pg. 33 7.5 Costs……………………………………………………………………….... Pg. 34 7.6 Institutional Role……………………………………………………………. Pg. 35 8. The Current State of RDM in Canada…………………………………...... Pg. 36 8.1 Gap Analysis……………………………………………………………….. Pg. 36 8.2 Readiness Assessment…………………………………………………… Pg. 40 Conclusion………………………………………………………………………. Pg. 43

3

List of Acronyms AHRC: Arts and Humanities Research Council (UK)

ANDS: Australian National Data Service

ARC: Australian Research Council

BBSRC Biotechnology and Biological Sciences Research Council (UK)

CARL: Canadian Association of Research Libraries

CARA: Canadian Association of Research Administrators

CFI: Canada Foundation for Innovation

CIHR: Canadian Institutes of Health Research

DCC: Digital Curation Centre (UK)

DMP: Data Management Plan

EC: European Commission

EPSRC: Engineering and Physical Sciences Research Council (UK)

ESRC: Economic and Social Research Council (UK)

IPY: International Polar Year

MRC: Medical Research Council (UK)

NERC: Natural Environment Research Council (UK)

NIH: National Institutes of Health (US)

NRC: National Research Council

NSERC: Natural Sciences and Engineering Research Council of Canada

NSF: National Science Foundation (UK)

OECD: Organization for Economic Cooperation and Development

RCUK: Research Councils UK

RDA: Research Data Alliance

RDC: Research Data Canada

RDM: Research Data Management and Sharing

REBs: Research Ethics Boards

SSHRC: Social Sciences and Humanities Research Council of Canada

STFC: Science and Technology Facilities Council (UK)

TCPS: Tri-Council Policy Statement on the Ethical Conduct for Research Involving Humans

4

1. Executive Summary The volume of scientific data around the world is increasing at a phenomenal rate. According to a report of the Canadian Research Data Summit in 2011, “the way that we choose to manage our research data will directly impact our ability to undertake leading edge research and development in the future.”1 But managing data is about much more than supporting research excellence. “Digital data are the raw materials of the knowledge economy and are becoming increasingly important for all areas of society. Policies, services and infrastructure must be in place if we are to capitalize on this rising tide of data.”2

This brief presents the results of an environmental scan undertaken in the fall/winter of 2014-2015. It provides an up-to-date and detailed overview of the policy environment for research data management and sharing in Canada and internationally. The report identifies some of the major challenges related to policy adoption and concludes with a readiness assessment for policy implementation in Canada.

The scan found that there are a growing number of research data management and sharing policies (herewith referred to as RDM policies) being adopted by funding agencies and institutions around the world. The objectives of these policies are to improve the efficiency of research, support the re-use of data for new insights and discoveries, foster collaboration, and facilitate greater transparency. To achieve these policy objectives, research data must be properly managed across the data lifecycle.

The jurisdictions with the most comprehensive policy environments are the United Kingdom, United States, Australia and European Union. Details of policies vary across regions, agencies and domains, but they also have a number of things in common. The most frequent policy components are requirements around standards and metadata, data sharing, and data retention and/or long-term preservation. Data management plans (DMPs) are usually required in the context of these policies, as they compel researchers to think about how they will manage their data in advance of the project, a key requisite for good data management practices. Policies also consistently contain provisions for the protection of confidentiality, intellectual property, and sensitive data.

There is significant diversity in terms of how policies are monitored and implemented. Research data management policies are new to many organizations and most are still working out how to administer them appropriately. In some cases, policies are adopted with little or no monitoring of compliance. In other instances, DMPs are attached to proposals and undergo a light review by peer review committees, but with little or no follow-up at the end of the project. Still in other cases, policy compliance is reviewed at the end of a project and there are consequences for non-compliance. Likewise, some organizations have chosen to phase in their policy. The European Commission, for example, has begun with a pilot project that requires 15% of their funded research projects to develop data management plans. The Engineering and Physical Sciences Research Council (EPSRC) in the UK has taken the distinctive approach of requiring universities (rather than researchers) to develop roadmaps that will ensure the Council’s policy can be adhered to by funded researchers. Regardless of how policies are managed or implemented, support and guidance for researchers is essential to ensure compliance, since many researchers are not familiar with what is involved in good data management practices. It is also clear that full adherence to any policy will take time and will likely happen incrementally. 1  http://www.rdc-­‐drc.ca/wp-­‐content/uploads/Report-­‐of-­‐the-­‐Canadian-­‐Research-­‐Data-­‐Summit1.pdf  2  http://www.rdc-­‐drc.ca/wp-­‐content/uploads/Report-­‐of-­‐the-­‐Canadian-­‐Research-­‐Data-­‐Summit1.pdf  

5

In Canada, the federal government has been increasing its interest and support for research data management and sharing through open government and open science initiatives. “Canada’s Science, Technology and Innovation Strategy 2014” includes a section promoting open science through the facilitation of “open access to publications and related data resulting from federally-funded research in order to accelerate research, drive innovation and benefit the economy”3. In February 2015, the agencies announced a new “Tri-Agency Open Access Policy on Publications” that requires research publications supported by public funds to be made openly available for the benefit of the community at large.4 The TC3 are currently assessing how to move forward with research data management within this broader policy context. Unquestionably, policies cannot be adopted in isolation. Good research data management practices will depend on multiple contributing factors including incentives, expertise, services and infrastructure, as well as appropriate funding mechanisms. This review found that the situation for RDM is improving and Canada has made significant progress since signing the “OECD Declaration on Access to Research Data from Public Funding” in 2004, with both bottom-up and top-down advances in RDM infrastructure, services and expertise.

That being said, Canada still lacks infrastructure, services and funding mechanisms to support widespread RDM. Infrastructure funding remains focused on domain-based solutions that support research excellence, rather than data sharing and preservation after the lifespan of the project. Portage, a national library-based network for managing research data (led by the Canadian Association of Research Libraries), and its collaborators (Research Data Canada and Compute Canada) are laying the foundation for more horizontal infrastructures and services, but this is a grassroots effort that will have difficulties expanding without external funding.

There are other challenges. Institutions and researchers still need to be convinced. While there are some research communities that have embraced a culture of data management and sharing, many stakeholders do not think RDM should be a priority for the research community. Both researchers and institutions are apprehensive about taking on greater responsibility for managing research data. Researchers are worried about the time, knowledge and resources involved in preparing data. Institutions are concerned about how they will fund data management support services and infrastructures. Parallel efforts must be made to increase acceptance of policy objectives within the various stakeholder communities, including the adoption of appropriate incentive schemes. Skills and expertise in the area of RDM must also be expanded.

Despite the challenges, it is clear that policies are an extremely powerful lever to push the community forward. They provide a framework that helps to guide best practices and without them it is unlikely that there will be widespread adoption of RDM in Canada. Countries that have chosen to move ahead with policy implementation have found that although full compliance cannot be expected immediately, policies can greatly assist in raising awareness of RDM. As noted in a 2013 TC3+ consultation document, “Canada now stands in direct competition with a host of other countries… in the race to develop an effective strategy for harnessing the digital wave.”5 RDM policies are an important component of any such strategy.

3 http://www.ic.gc.ca/eic/site/icgc.nsf/eng/07482.html#promoting 4 http://www.science.gc.ca/default.asp?lang=En&n=F6765465-1  5 http://www.sshrc-crsh.gc.ca/about-au_sujet/publications/digital_scholarship_consultation_e.pdf

6

2. Introduction Over the past ten years we’ve witnessed the acceleration of a significant worldwide trend towards research data sharing and management. This trend is progressing in parallel with a movement calling for open access to research publications and can be situated within a broader effort to ensure that the results of publicly funded research are available to the public. In 2004, 34 countries including Canada signed the “OECD Declaration on Access to Research Data from Public Funding”.6 Through this declaration, signatories were recognizing that open access to, and unrestricted use of, data promotes scientific progress and contributes to new discoveries and innovation. In addition, they were acknowledging that data sharing maximizes the value derived from public investments and supports re-use of data across disciplinary and jurisdictional boundaries. Since that time, the momentum for data sharing and management has grown and investments in research data management policies and services have increased dramatically.

At the same time, in Canada the federal government has been increasing its commitment to open science. This originated with open government and open data initiatives, and has grown in importance with the adoption of the “Open Data Charter” in June 2013 at the G8 (now G7) Summit in Lough Erne, Northern Ireland. At this summit, Canada and all other G8 members agreed to implement a set of open data principles and best practices that would lay the foundation for the release and reuse of government data before December 31, 2015. 7 “Canada’s Science, Technology and Innovation Strategy 2014” identified open science as a priority, as has the Open Government Initiative. In order to deliver on these commitments, the Government of Canada has prepared an Open Government Action Plan with three streams: open access, open data, and public engagement.

In parallel, the three federal granting agencies have been examining the issues around open access and research data management. In 2011, they commissioned an environmental scan of the policy context related to open access and research data management. The results were presented in a “Brief on Open Access to Publications and Research Data”8, which provided background information about both the growing trend towards open access to publications as well as sharing of research data. In 2013 the TC3+ (CIHR, NSERC, SSHRC, CFI) and Genome Canada undertook a consultation with the stakeholder community to further engage and develop a strategy for a more comprehensive and coordinated approach to RDM in Canada. And most recently, in February 2015, the agencies released a “Tri-Agency Open Access Policy on Publications” that requires research publications supported by public funds to be made openly available for the benefit of the community at large.9

The aim of this brief is to inform the community about the state of RDM policy development internationally and in Canada. This brief presents the results of an environmental scan undertaken in the fall/winter 2014-2015. It provides an up-to-date and detailed overview of the RDM policy environment. The report discusses some of the major challenges related to policy adoption, and concludes with a gap analysis and readiness assessment for policy implementation in Canada.

6 http://acts.oecd.org/Instruments/ShowInstrumentView.aspx?InstrumentID=157 7 http://data.gc.ca/eng/g8-open-data-charter-canadas-action-plan 8 http://science.gc.ca/default.asp?lang=en&n=2360F10C-1  9 http://www.science.gc.ca/default.asp?lang=En&n=F6765465-1  

7

3. Policy Environment While certainly not ubiquitous, there are a growing number of research data management and/or sharing policies being adopted by funding agencies and institutions around the world. The purpose of these policies is to improve data management practices in order to allow data produced through research to be shared and re-used by others, enable the verification of research results, and ultimately fuel further innovation.

3.1 Policy Objectives and Principles Research data management policies are situated in the context of a broader set of principles and objectives that guide their specific requirements. In general, policies will support several (or all) of the objectives listed below:

• Accelerate research • Support new insights and discoveries • Foster collaboration • Improve efficiency of research • Facilitate accountability

The 2003 OECD declaration, of which Canada was a signatory, outlines a comprehensive set of principles for RDM that have informed the principles adopted by many other organizations and remain relevant over a decade later10:

• Openness: balancing the interests of open access to data to increase the quality and efficiency of research and innovation with the need for restriction of access in some instances to protect social, scientific and economic interests

• Transparency: making information on data-producing organizations, documentation on the data they produce and specifications of conditions attached to the use of these data, available and accessible internationally

• Legal conformity: paying due attention, in the design of access regimes for digital research data, to national legal requirements concerning national security, privacy and trade secrets

• Formal responsibility: promoting explicit, formal institutional rules on the responsibilities of the various parties involved in data-related activities pertaining to authorship, producer credits, ownership, usage restrictions, financial arrangements, ethical rules, licensing terms, and liability

• Professionalism: building institutional rules for the management of digital research data based on the relevant professional standards and values embodied in the codes of conduct of the scientific communities involved

• Protection of intellectual property: describing ways to obtain open access under the different legal regimes of copyright or other intellectual property law applicable to databases as well as trade secrets

10http://acts.oecd.org/Instruments/ShowInstrumentView.aspx?InstrumentID=157  

8

• Interoperability: paying due attention to the relevant international standard requirements for use in multiple ways, in co-operation with other international organizations.

• Quality and security: describing good practices for methods, techniques and instruments employed in the collection, dissemination and accessible archiving of data to enable quality control by peer review and other means of safeguarding authenticity, originality, integrity, security and establishing liability

• Efficiency: promoting further cost effectiveness within the global science system by describing good practices in data management and specialized support services

• Accountability: evaluating the performance of data access regimes to maximize the support for open access among the scientific community and society at large

3.2 Typical Policy Requirements The features of a given RDM policy will reflect the particular objectives and principles on which it is based. Therefore, while many policies contain similar elements, there may be greater emphasis on some requirements over others. For example, a policy based on the principle of data sharing will likely concentrate on key practices needed for providing access to the data, while a policy based on data stewardship will focus on the roles and responsibilities involved in managing data.

The most common elements of the RDM policies reviewed for this scan are outlined in the table below:

Table 1: Common elements of a RDM policy Policy requirements

Data quality and standards

Investigators are required to adhere to international standards to enable access and reuse.

Data documentation and metadata must accompany data so that the data are understandable by others.

Data access and sharing

Investigators are required to make data available to be shared (usually upon publication of results or shortly thereafter, although some agencies do allow embargo periods).

Requirements for deposit of metadata into a local or national catalogue.

Data retention and preservation

Data are required to be retained for a certain minimum time period. Where possible, investigators must deposit their data in a long-term archive to ensure the preservation of their data.

Data management plans

Research proposals must include a Data Management Plan.

9

Common provisions to policies

Privacy The rights and privacy of individuals who participate in research must be protected at all times. Thus, data made available for broader use should be free of identifiers that would permit linkages to individual research participants and variables that could lead to deductive disclosure of the identity of individual subjects.

Traditional knowledge Where local and traditional knowledge is concerned, rights of the knowledge holders shall not be compromised.

Data of a sensitive nature

Where data release may cause harm, specific aspects of the data may need to be protected (for example, locations of nests of endangered birds or locations of sacred sites).

Intellectual property/Data ownership

It may be necessary on occasion to delay publication for a short period to allow time for applications to be drafted.

Other aspects

Principles Data policies adhere to a set of overarching principles that articulate their value.

Scope/Coverage of Policy

Describe the scope of data covered by the policy.

Roles and responsibilities

The policy identifies the various parties responsible for managing data across the different stages of the lifecycle.

Monitoring and enforcement

The means by which policies will be monitored or enforced are outlined in the policy.

3.3 International The jurisdictions that are most advanced in terms of research funders’ policies are the United Kingdom and the United States. The different agency policies vary across organizations in terms of their strength, coverage, roles and responsibilities, and requirements. A written summary of the major funders’ policies follows.

United Kingdom: In 2011, Research Councils UK (RCUK) issued a set of “Common Principles on Data Policy”. The principles call for data to be made openly available with as few restrictions as possible in a timely and responsible manner. Since then, each of the seven councils in the Research Council UK has implemented a policy on access to research data, as has the Wellcome Trust (a major charitable organization that funds biomedical research). The UK funders’ policies range in terms of their requirements and details, but they are generally aligned with the Common Principles. According to an overview published by the University of Bath11, policies typically cover the following elements:

• Types of data covered by the policy 11 http://www.bath.ac.uk/research/data/policy/funder-data-policies.html

10

• Expectations for data sharing including access and timescales

• Minimum data retention periods

• Use of metadata and documentation standards

• Justified exemptions to data sharing

• Costs associated with data management that may be paid for through grants

• Requirements for submission of data management plans with grant applications

• Acknowledgement of data creators

The following table of UK funders’ requirements was developed by the (DCC) Centre in the UK and provides a useful overview:

Table 2: DCC’s Overview of UK Funders’ Data Policies12 (A list of acronyms is provided at the beginning of the report)

The table illustrates that most of the UK funders require researchers to complete a data management plan; ensure that they are using appropriate metadata and standards; and retain data or deposit into a repository when available. For example, the Economic and Social Research Council requires researchers to prepare a data management plan and stipulates, “[t]he data must be made available for re-use or archiving with the ESRC data service providers within three months of the end of the grant.”13

Alternatively, the Arts and Humanities Research Council (AHRC) requires a technical plan in cases where digital outputs or digital technologies are an essential part to the planned research outcomes. The plan should give a summary of those outputs, explain the technical methodology, technical support and relevant experience, and address preservation, sustainability and use. The AHRC also requires that significant electronic

12 http://www.dcc.ac.uk/resources/policy-and-legal/overview-funders-data-policies  13 http://www.esrc.ac.uk/about-esrc/information/data-policy.aspx

11

resources or datasets be made available in an accessible depository for at least three years after the end of the grant.14

The Engineering and Physical Sciences Research Council (EPSRC) has taken a somewhat different approach from the other councils. In its policy, EPSRC has set out clear expectations for institutions they fund, which include a requirement that institutions develop of an institutional ‘Roadmap’. This is discussed in more detail below, in the section entitled Approaches to Policy Implementation.

United States: In order to improve the management of research data produced through publicly funded research, the White House’s Office of Science and Technology Policy (OSTP) published a policy memorandum that directed all federal agencies with more than $100M in R&D expenditures to require researchers to better account for and manage the digital data resulting from federally funded scientific research. Each of the 22 agencies subject to these requirements was required to develop a plan that outlines how they will adhere to this policy. Plans for fulfilling this directive are being developed by all agencies and are beginning to be made publicly available. Several agencies including the Department of Defence, Department of Energy, NASA, National Science Foundation (NSF), and the National Institute of Standards and Technology have all released plans (or draft plans) that will require funded researchers to develop data management plans.15

Both the National Institutes of Health (NIH) and the NSF implemented RDM policies before the OSTP directive. Adopted in 2001, NIH was one of the first funding agencies to have a policy about research data sharing. The policy states “[d]ata should be made as widely and freely available as possible while safeguarding the privacy of participants, and protecting confidential and proprietary data.”16 In addition, investigators submitting a research application requesting $500,000 or more of direct costs in any single year to NIH are expected to include a plan to explain how they will share their data, or explain why data sharing is not possible.

The NSF policy states, “[i]nvestigators are expected to share with other researchers, at no more than incremental cost and within a reasonable time, the primary data, samples, physical collections and other supporting materials created or gathered in the course of work under NSF grants. Grantees are expected to encourage and facilitate such sharing.”17 In 2011, the NSF introduced a new requirement that all proposals must include a supplementary document of no more than two pages labeled “Data Management Plan”. The plan should describe how research teams will conform to the policy. On March 2015, NSF released their plan for pubic access to NSF funded research as requested by OSTP, which underscored their ongoing support for data management plans.18 The plan also encourages researchers to cite their data in the context of publications and proposals.

European Commission: In Horizon 2020, the European Commission’s (EC) financial instrument for funding research from 2014-2020, a pilot action on open access to research data will be implemented with the aim of improving and maximizing access to 14 http://www.dcc.ac.uk/resources/policy-and-legal/research-funding-policies/ahrc  15 http://science.energy.gov/funding-opportunities/digital-data-management/ 16 http://grants.nih.gov/grants/policy/data_sharing/data_sharing_guidance.htm 17 http://www.nsf.gov/bfa/dias/policy/dmp.jsp 18 www.nsf.gov/news/special_reports/public_access/

12

and re-use of research data generated by funded projects. The “Pilot on Open Research Data” will be monitored throughout Horizon 2020 with a view to further developing EC policy on open research. As part of the pilot, projects that fall into 7 pre-selected research areas19 will be required to develop a data management plan outlining how they will manage and provide access to their data. Pilot areas represent about 20% of the overall research funding budget of Horizon 2020. Projects participating in the pilot must also deposit data into a research data repository and take measures to make it possible for third parties to access, mine, exploit, reproduce and disseminate — free of charge for any user — both the data needed to validate results and any other data generated in the project.20 The EC does not currently specify what this would entail, but presumably they will be asking researchers to assign re-use licenses to their data when depositing. More information about this project is described in a later section, Approaches to Policy Implementation. Australia: Developed jointly by the National Health and Medical Research Council, the Australian Research Council and Universities Australia in 2007, “The Code for the Responsible Conduct of Research” contains a section on the “Management of Research Data and Primary Materials”. Primary materials are defined as objects (physical or virtual) acquired through a process of scholarly investigation from which Research Data may be derived. 21 The Code outlines the roles and responsibilities of different stakeholders in the research process, and assigns responsibility for the management and retention of research data to both researchers and institutions. It requires institutions to have policies on the retention and management of materials and research data, stating: “It is important that institutions acknowledge their continuing role in the management of research material and data. The institutional policy must be consistent with practices in the discipline, relevant legislation, codes and guidelines.” It goes on to say, “[i]nstitutions must provide facilities for the safe and secure storage of research data and for maintaining records of where research data are stored.”22 The Code directs researchers to manage their research data and primary materials according to their institutional policy. Although the Code is not stringently applied nor enforced by ARC, it has been an incentive for improving RDM practices in Australia and a number of universities have developed RDM policies.

Despite the lack of strong policies, Australia is still considered one of the leaders in RDM in that it has made major investments in its services and infrastructure. In 2007, the Australian Government through the National Collaborative Research Infrastructure Strategy Program created the Australian National Data Service (ANDS). ANDS invests in and hosts a number of local and centralized services, including Research Data Australia, a national discovery service to promote visibility of research data collections.

Germany: The German Research Foundation (DFG), one of the major scientific funding agencies in Germany includes a section about “data handling” in all grant proposals. This section asks researchers to describe if and how data will be made available for future reuse by other researchers. 23 Researchers can request funding for making research data available for future reuse, but they must also describe how the institutions

19 http://europa.eu/rapid/press-release_IP-13-1257_en.htm 20 http://ec.europa.eu/research/participants/data/ref/h2020/grants_manual/hi/oa_pilot/h2020-hi-oa-pilot-guide_en.pdf 21 http://ppl.app.uq.edu.au/content/4.20.06-research-data-management 22 http://www.nhmrc.gov.au/_files_nhmrc/publications/attachments/r39.pdf 23 http://www.dfg.de/formulare/54_01/54_01_en.pdf

13

participating in the project will contribute to data and information management.”24 In addition, in 2010, the Alliance of German Science Organizations adopted a set of principles for the handling of research data25.

Others: Funders and research organizations in Chile26 and Finland27 are also in the process of developing RDM policies in their jurisdictions. The Fundação para a Ciência y a Tecnologia in Portugal recommends, but does not (yet) require, that researchers applying for funding share the primary data and other materials produced in projects financed by FCT with other researchers and include data management plans in their funding applications.28

3.4 Canada In 2010, the three federal granting agencies released a set of principles relating to access to research outputs, which states the following: “CIHR, NSERC and SSHRC are committed to developing a shared approach for improving access to publicly funded research in keeping with internationally recognized best practices, standards and policies for funding and conducting research. The following principles will guide the agencies in promoting access to research results:

• Advance Knowledge: The advancement of knowledge depends upon peer review to ensure excellence, as well as long-term preservation to ensure that research can be built upon by others

• Minimize Research Duplication: Broad dissemination increases the effectiveness of public investments in research by reducing the potential for unnecessary duplication

• Maximize Research Benefits: Publicly funded research should be as accessible as possible in order to maximize the economic, social, cultural and health benefits for Canadians

• Promote Research Accomplishments: Improving access to research results will better promote the accomplishments of Canadian researchers throughout the world”29

These principles are consistent with other international statements about research data management and sharing such as the 2004 OECD Declaration and the 2013, G8 Science Ministers’ Statement30 addressing open scientific research data.

As part of these broader policy objectives to improve access to the results of research and to increase the dissemination and exchange of research results, the TC3 announced a new “Tri-Agency Open Access Policy on Publications”31 in early 2015. This policy requires all grant recipients to ensure that any peer-reviewed journal publications

24 Ibid 25 www.allianzinitiative.de/en/core_activities/research_data/principles 26 http://datoscientificos.cl/files/manual-2014.pdf 27 http://www.aka.fi/en-GB/A/Funding-and-guidance/How-to-apply/Appendices/Research-plan 28 https://www.fct.pt/documentos/PoliticaAcessoAberto_Dados.pdf (translated in English by automated translating tool) 29 http://www.science.gc.ca/default.asp?Lang=En&n=9990CB6B-1 30 https://www.gov.uk/government/news/g8-science-ministers-statement 31 http://www.science.gc.ca/default.asp?lang=En&n=F6765465-1

14

arising from Agency-supported research are freely accessible within 12 months of publication.

In terms of data management, the 2011 “Tri-Agencies Framework: Responsible Conduct of Research” 32 outlines the appropriate conduct of research. Section 2.1.2 of the framework states that at minimum, researchers are responsible for:

a. Using a high level of rigour in proposing and performing research; in recording, analyzing, and interpreting data; and in reporting and publishing data and findings

b. Keeping complete and accurate records of data, methodologies and findings, including graphs and images, in accordance with the applicable funding agreement, institutional policies and/or laws, regulations, and professional or disciplinary standards in a manner that will allow verification or replication of the work by others

Several Canadian research funders have also adopted more explicit RDM policies. CIHR requires all grant recipients to retain original data sets arising from CIHR-funded research for a minimum of five years after the end of the grant. This applies to all data, whether published or not. In addition, for research in some areas (bioinformatics, atomic, and molecular coordinate data), data must be deposited into public repositories.33

SSHRC has had a “Research Data Archiving Policy” in place since 1990. The policy states, “[a]ll research data collected with the use of SSHRC funds must be preserved and made available for use by others within a reasonable period of time. SSHRC considers ‘a reasonable period’ to be within two years of the completion of the research project for which the data was collected.” However, SSHRC has not actively enforced the policy and few SSHRC funded researchers are aware of it.

In 2008, Genome Canada adopted a “Data Research and Resource Sharing Policy”. The policy “expects data to be released and shared no later than the original publication date of the main findings from any datasets generated by that project.”34 In addition, applicants must provide a Data and Resource Sharing Plan as part of their application.

NSERC does not have an RDM policy, but it has implemented requirements in the context of specific programs including International Polar Year and the Discovery Frontiers (jointly with Genome, CIHR, and CFI). The International Polar Year (IPY) was a large scientific program focused on research in the Arctic and the Antarctic regions from March 2007 to March 2009. With a budget of $1.2 billion US, IPY involved more than 60 countries, over 200 international research networks, and thousands of researchers. In order to meet its objectives of interdisciplinary and international collaboration and to ensure a lasting legacy, IPY committed to ensuring full, free, and open access to IPY data as described in the IPY Data Policy.35 This was one of the most comprehensive policies for research data at that time and remains so to date.

The Heart and Stroke Foundation, which provided $38 million in funding to 1,500 researchers in 2013, has a data policy that requires grant recipients to deposit

32 http://www.rcr.ethics.gc.ca/eng/policy-politique/framework-cadre/  33 http://www.cihr-irsc.gc.ca/e/46068.html#5.1.2 34 http://www.genomecanada.ca/medias/PDF/EN/DataReleaseandResourceSharingPolicy.pdf 35 http://www.api-ipy.gc.ca/pg_IPYAPI_050-eng.html

15

bioinformatic, atomic, and molecular coordinate data into the appropriate public database immediately upon publication of research results.36

The following table compares the different elements of research data management policies at some of the major agencies reviewed for this scan.

Table 3: Comparison Table of Funders’ Policies

Agency Domains Coverage Timing for data sharing

Monitoring DMPs Specified Repository

Funds available for RDM

Australia-ARC All All Not specified

No No No Not specified

Canada-CIHR Health All Upon publication of research results

No No bioinformatics, atomic, and molecular coordinate data into public databases

No

Genome Canada

Genomics All Upon publication

Yes Yes Yes Not specified

Canada-Heart and Stroke Foundation

Life sciences

All Upon publication

? No Yes- when a repository is available in a discipline

Not specified

Canada-SSHRC

Humanities and social sciences

All Within two years of the completion of the research project for which the data was collected

No No Institutional or domain repository

Not specified

European Commission- Horizon 2020

All Selected areas

Not Specified

Yes Pilot project with specific disciplines- opt out available

Available repository

Yes

Portugal- All All, but voluntary

Not specified

No Yes No Not specified

UK- AHRC Arts and humanities

All At least within three years after the end of the grant

No Yes No Yes

36 http://www.hsf.ca/research/en/hsf-open-access-research-outputs-policy-guidelines

16

Agency Domains Coverage Timing for data sharing

Monitoring DMPs Specified Repository

Funds available for RDM

UK- BBSRC Biotechnology and biological sciences

All No later than the release of main findings through publication

Yes Yes In specific disciplines

Yes

UK- CRUK Health (cancer research)

All no later than the acceptance for publication of the main findings

Yes Yes No No

UK- EPSRC Engineering and physical sciences

All Not specified

Institutions must develop a roadmap

No Yes- Institutional responsibility

No

UK- ESRC Economic and Social Research

All At or around the time of publication

Yes Yes Yes-UK Data service

Yes

UK- MRC Medical All Not specified

Yes Yes No Yes

UK- NERC Environmental science

All No later than of two years from the end of data collection

Yes Yes Yes-NERC data centres

Yes

UK- STFC Science and technology

All Not specified

Review of DMPs

Yes No Not specified

UK- Wellcome Trust

Biomedical sciences

data holding significant value as a resource for the wider research community

Upon publication of their research

End of grant report

Yes Yes- discipline and institutional repositories

Yes

US- NIH Health All No later than the acceptance for publication of the main findings from the final data set

Yes For grants that exceed 500k/year

No Yes

US- NSF Sciences and Engineering

All within a reasonable time

Yes Yes No Yes

17

3.5 Other Stakeholders Funders are not the only actors developing policies related to research data management and sharing. Journals, projects and institutions are also adopting policies. Typical policy elements of these actors mirror those of funders, with variations based on community standards and practices, as well as the availability of repositories.

Some journals, mainly in the life sciences, require researchers to deposit any data available related to their articles into a repository for verification. These policies typically require authors to make their data available to others, and where public repositories exist, that authors deposit their data into these repositories. There is some indication that journal data policies may soon expand beyond the life sciences. PLOS, for example, a major publisher in the sciences and medicine, has recently implemented a policy for all of its journals requiring “authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception.”37 That being said, a recent review of the current state of journal sharing policies found that there is still a large percentage of journals that do not have policies on data sharing, and where policies exist, they are often vague and rarely enforced.38

A growing number of research projects are also implementing RDM policies. These policies are most common in the context of international projects, in which it is important to establish common approaches across jurisdictions. The International Polar Year is one example, as is the International Barcode of Life (iBOL) project. iBOL has a data release and resources sharing policy, which “seeks to accelerate the timely development of products that will benefit humankind by providing rapid access to the primary outputs from iBOL: DNA sequences associated with high quality meta-data including taxonomic assignments. The working philosophy of the iBOL project is full release of data within 18 months of a sequence being produced. There is the expectation that this 18-month time period will be reduced as the project progresses, and from the outset more rapid data release is encouraged whenever practical.”39

Ocean Networks Canada has a data policy that states, “[d]ata collected by Ocean Networks Canada are for research and education purposes (not for profit), and are generally open access and free to anyone. Ownership of the data lies with the instrument owner, which in most cases is Ocean Networks Canada.”40 The policy does contain exceptions to availability for preliminary data or in cases where data will be used for commercial purposes.

Universities have also begun to implement RDM policies. Policies are most prolific in the UK41 and Australia42, likely because both countries have funding agencies that have placed some responsibility for RDM with the institution. According to the Digital Curation Centre, there are currently 20 UK universities with an RDM policy, and 6 that have draft

37 http://www.plosone.org/static/policies.action#sharing 38 Sturges, Paul and Bamkin, Marianne and Anders, Jane H.S. and Hubbard, Bill and Hussain, Azhar and Heeley, Melanie (2014) Research data sharing: developing a stakeholder-driven model for journal policies. Journal of the Association for Information Science and Technology. ISSN 2330-1643 (Available at: http://eprints.nottingham.ac.uk/3185/) 39 http://ibol.org/resources/data-release-policy/ 40 http://www.oceannetworks.ca/data-tools/data-help/data-policy 41 http://www.dcc.ac.uk/resources/policy-and-legal/epsrc-institutional-roadmaps 42 http://www.ands.org.au/datamanagement/policy.html

18

policies in development. 43 The Australian National Data Service currently lists 5 institutional policies in Australia.44

University RDM policies tend to contain similar elements as those found in funders’ policies, while also defining the specific role for the institution. The University of Edinburgh provides an example of a very comprehensive university policy that could act as an exemplar for other institutions. 45 Through this policy, the institution accepts responsibility for tracking, preserving and supporting researchers in managing their data. The policy requires faculty to develop data management plans. It also commits the university to providing support, training and “mechanisms” for storage and sharing. The university acknowledges that it is an aspirational policy and will take some years to implement fully. Clearly, even the most advanced universities in terms of RDM services and infrastructure would not assume that all researchers are able to adhere to requirements, and policies must still be phased-in.

Many institutions have policies on ethical conduct that require the proper handling of data by the researchers including accurate presentation and retention of data. Two Canadian universities also have more specific policy requirements addressing data management practices. The University of Alberta promotes the use of data management plans (DMPs) in the context of the University's policy on the stewardship of research records. The University’s “Research Records Stewardship Guidance Procedure” makes two references to Data Management Plans: “The articulation of the primary stewardship responsibilities for all parties throughout the research lifecycle should be made at the very beginning of a research project in a Data Management Plan,” and “[w]ith regard to human participant research generally, records do not have to be destroyed, provided the researcher’s Data Management Plan has a clear statement about appropriate records management, storage and retention.”46

The University of PEI has a policy on “Open Access and Dissemination of Research Output”, which encourages the deposit of research data into the UPEI Virtual Research Environment (VRE). It asserts that, “research data be made accessible in a fashion and timeline deemed appropriate by the researcher/research group. Where possible, research data would be made publicly accessible on publication of results of the research. Where privacy rights of human subjects conflicts with full public access, the researcher/research group will aim for the most public access possible and consistent with privacy, for example by providing anonymized data, or providing full access to data to other research groups that can demonstrate having met acceptable research ethics guidelines for handling such private information.”47

43 http://www.dcc.ac.uk/resources/policy-and-legal/institutional-data-policies 44 http://www.ands.org.au/datamanagement/policy.html 45 http://www.ed.ac.uk/schools-departments/information-services/about/policies-and-regulations/research-data-policy 46 https://policiesonline.ualberta.ca/PoliciesProcedures/Procedures/Research-Records-Stewardship-Guidance-Procedure.pdf 47 https://cab.upei.ca/sites/default/files/attachments/OpenAccessandDisseminationofResearchOutput.pdf

19

4. Data Management Plans Increasingly, many research funders require a data management plan (DMP) as a component of funding applications. DMPs are formal documents that “typically state what data will be created and how, and outline the plans for sharing and preservation, noting what is appropriate given the nature of the data and any restrictions that may need to be applied.” 48 DMPs are seen as a way of improving data management practices during the research process and they compel researchers to establish how they plan manage their data in advance of a project. Writing a DMP helps organize the research process and provides consistent guidelines for handling data, making the research process more efficient. In addition, DMPs can reduce the costs of research, as early planning for research data management has been shown to significantly reduce costs of data management over the long term.49

In all the examples reviewed for this scan, DMPs represent only one element of a broader research data management policy. The policy provides the requirements and the DMP outlines how requirements will be adhered to. From a policy perspective, DMPs are an important tool for ensuring that researchers are aware and have a plan to adhere to policy requirements in advance of starting to collect their data.

Common requirements for DMPs have been outlined in a checklist developed by the DCC in the UK and are documented below. The emphasis on different elements will vary and often depend on the focus of the policy requiring the DMP. For example, a policy focus on data sharing may emphasize the data sharing elements of the policy over other elements.

Table 4: DCC Checklist for Data Management Plans

Data Collection What data will be collected or created?

How will the data be collected or created?

Documentation and Metadata

What standards, documentation and metadata will accompany the data?

Ethics and Legal Compliance

How will ethical issues be managed?

How will copyright and Intellectual Property Rights (IPR) issues be managed?

Storage and Backup How will the data be stored and backed up during the research?

How will access and security be managed?

Retention and Preservation

Which data should be retained and/or preserved?

What is the long-term preservation plan for the data?

Data Sharing How will the data be shared?

Are any restrictions on data sharing required?

48 http://www.dcc.ac.uk/resources/data-management-plans 49 http://ukdataservice.ac.uk/manage-data/plan/costing.aspx  

20

Responsibilities and Resources

Who will be responsible for data management?

What resources will be required to deliver the data management plan?

In practice, DMP requirements vary significantly across organizations. Most of the differences lie in the level of detail required in the plan and the types of guidance and support offered. Some notable areas of variation are described below.

Level of detail: Some organizations require very comprehensive descriptions across all elements listed above, while others ask for much less detail. In addition, some organizations are very prescriptive about how the DMP should be formatted, while others leave it up the individual researchers to include whatever information they think is relevant. The trend seems to be towards greater specificity of requirements, as this helps to guide researchers in terms of what a data management plan should entail.

Timing and versioning of DMPs: In the majority of cases, DMPs must be completed and attached to the funding proposal in advance of the project launching. However, there are a few exceptions. In the Horizon2020 Pilot Project for example, the European Commission expects the first version of the DMP to be delivered within the first 6 months of the project, with more elaborated versions delivered at later stages of the project. The EPSRC in the UK requires data management plans, but does not review them. DMPs may change over time, and some organizations (like Horizon2020) ask researchers to update their DMPs regularly if there are changes to the original plan.

Scope: The scope of the data to be addressed in the DMP also varies across funders. Some policies target only the research data that underpins the publications. Others, such as the Wellcome Trust, require a DMP only when the research ”involves the generation of datasets that have clear scope for wider research use and hold significant long-term value.”50 Still others require that all data produced in a project be made available, for example, the NIH requests plans for “all data from funded research that can be shared without compromising individual subjects' rights and privacy, regardless of whether the data have been used in a publication.”51

4.1 Sample Data Management Plans In order to demonstrate the different approaches to DMPs, a number of examples are offered below.

Economic and Social Research Council (UK): 52 All ESRC grant applicants planning to create data during their research have to include a data management plan with their application. A data management plan helps to decide how research data will be managed throughout the research cycle and will be available for sharing afterwards. Most research data can be successfully archived and shared.

ESRC expects award holders to consider all issues related to confidentiality, ethics, security and copyright before initiating the research. Any challenges to data sharing (e.g. copyright or data confidentiality) should be critically considered in a plan, with possible solutions discussed to optimize data sharing.

50 http://www.wellcome.ac.uk/About-us/Policy/Spotlight-issues/Data-sharing/Guidance-for-researchers/index.htm#_B._When_is_a%20data%20management%20and%20sh 51 http://grants.nih.gov/grants/policy/data_sharing/data_sharing_faqs.htm#901 52 http://ukdataservice.ac.uk/manage-data/plan/dmp-esrc.aspx  

21

A data management plan includes the following topics:

• Assessment of existing data • Information on new data • Quality assurance of data • Backup and security of data • Expected difficulties in data sharing • Copyright/Intellectual Property Rights • Responsibilities • Preparation of data for sharing and archiving

Detailed guidance about preparing data management plans is provided by ESRC through the UK Data Archive.

Engineering and Physical Science Research Council (UK): The EPSRC does not require researchers to submit data management or sharing plans in grant applications. However, it does expect policies and plans to be in place.53 In clarifying its policy, the EPSRC states, “[i]t is suggested that research offices ensure appropriate provision for research data management is included in a research proposal before it is submitted to EPSRC. In particular: a) does a data management plan (DMP) exist? (EPSRC does not require DMPs with research grant applications, but our research data principles include that “…project specific data management policies and plans… …should exist for all data’)”54

Medical Research Council (UK): 55 All applicants submitting funding proposals to the MRC are required to include a DMP as an integral part of the application. The council asserts that everyone in a research team should have a clear sense of their responsibilities. Specific elements of the data management plan outlined in a template provided by MRC are as follows:

• Description of the data • Data collection / generation • Data management, documentation and curation • Data security and confidentiality of potentially disclosive information • Data sharing and access

Detailed guidance is also available to researchers if needed.

National Institutes of Health (US): Starting with the October 1, 2003 receipt date, investigators submitting an NIH application seeking $500,000 or more in direct costs in any single year are expected to include a plan for data sharing or state why data sharing is not possible. NIH is less prescriptive about the contents of data management plans, stating “[t]he precise content and level of detail to be included in a data-sharing plan depends on several factors, such as whether or not the investigator is planning to share data, the size and complexity of the dataset, and the like.”56

It then provides a number of examples of different types of data management plans to assist researchers in developing their own. However, NIH states that the plans should ideally cover the following elements:57

53 http://www.dcc.ac.uk/resources/policy-and-legal/research-funding-policies/epsrc 54 http://www.epsrc.ac.uk/files/aboutus/standards/clarificationsofexpectationsresearchdatamanagement/ 55 http://www.mrc.ac.uk/research/research-policy-ethics/data-sharing/data-management-plans/ 56 http://grants.nih.gov/grants/policy/data_sharing/data_sharing_guidance.htm#ex 57 http://grants.nih.gov/grants/sharing_key_elements_data_sharing_plan.pdf

22

• What data will be shared • Who will have access to the data • Where will the data be available • When will the data be shared • How will researchers locate and access the data

National Science Foundation (US): Each proposal must include a supplementary document of no more than two pages labeled “Data Management Plan”. This supplementary document should describe how the proposal will conform to NSF policy on the dissemination and sharing of research results.

The data management plan should include the following information:58

1. Products of the Research: The types of data, samples, physical collections, software, curriculum materials, and other materials to be produced in the course of the project

2. Data Formats: The standards to be used for data and metadata format and content (where existing standards are absent or deemed inadequate, this should be documented along with any proposed solutions or remedies)

3. Access to Data and Data Sharing Practices and Policies: Policies for access and sharing including provisions for appropriate protection of privacy, confidentiality, security, intellectual property, or other rights or requirements

4. Policies for Re-Use, Re-Distribution, and Production of Derivatives

5. Archiving of Data: Plans for archiving data, samples, and other research products, and for preservation of access to them

6. Certain directorates within the NSF, however, provide explicit guidelines and advice on forming data management plans, which may require more details59

5. Administering Policies Funders with RDM policies have taken a variety of approaches to administering them. Some of the key issues and current practices in terms of policy administration, such as guidance, evaluating DMPs, monitoring compliance, confidentiality and IP, and dealing with multiple policies are discussed here.

5.1 Policy Guidance Clear and detailed guidance on how to adhere to a data policy is essential to ensure compliance. However, given the intricacies of research data management, guidance can also become so complex that it is confusing for users. University of Edinburgh was one of the first UK universities to provide online research data management guidance in 2009. The resource aims to assist university researchers in complying with the increasingly demanding requirements of both external funding bodies and the university, and direct them to appropriate sources of support. Although well received, the guide was 58 http://www.nsf.gov/pubs/policydocs/pappguide/nsf11001/gpg_2.jsp#IIC2j 59 http://www.arl.org/focus-areas/e-research/data-access-management-and-sharing/nsf-data-sharing-policy/243-resources-for-data-management-planning#.VGTXGYd7Sqk

23

considered overly complex and had to be re-vamped. The new, much briefer version consists of eight pages covering the essential topics researchers must understand before embarking on a research project.60

In the US context, both NIH and NSF have recognized that guidance about adhering to their policy is critical, but also a challenge because practices and infrastructures vary significantly across disciplines and sub-disciplines. In order to navigate this, NSF has taken the approach of allowing the disciplines themselves to define best practices around managing data, and relies heavily on input from these communities for guidance and peer-review. It has been reported that, at least in the initial stages of this process, some disciplines are beginning to develop valuable resources to which researchers can turn for guidance.61

Given the complexities of RDM, some organizations have opted to provide access to experts who can give individualized support for RDM to researchers. For example, the Medical Research Council in the UK is developing a Data Support Service to facilitate and support data sharing for population and patient studies, in order to optimize the long-term use of rich data assets for new science. The project works closely with MRC data managers to coordinate and promote work on data sharing tools and standards and to promote the exchange of good practice. The Digital Curation Centre also provides numerous resources as well as consulting services to institutions across the UK that are developing RDM support services for their communities.

5.2 Evaluating DMPs Funders have also taken different approaches to evaluating DMPs. According to the literature, most of the UK funding councils are assessing DMPs during the initial peer review process. The Economic and Social Research Council (ESRC), for example, says that it “will seek an assessment of data management plans via its peer review and assessment processes. Although the application will first and foremost be assessed on grounds of its scientific merit, nonetheless, an assessment of the data management and sharing plan will be included in the general assessment of the application. ”62 A poorly prepared DMP, it goes on to explain, may have a detrimental effect on an otherwise strong application.

The Biotechnology and Biological Sciences Research Council (BBSRC) evaluates DMPs separately from the scientific excellence of the proposed research, “however, an application’s credibility will suffer if peer review agrees the statement is inappropriate. In the case where a highly rated proposal has an inappropriate Data Management Plan, Committees and Panels may choose to offer conditional awards and/or provide specific feedback to the applicants.”63

Genome Canada states that staff and review committees “will review the applicant’s proposed data and resource sharing plan to verify that it conforms to the Genome Canada policy and funds will not flow until an acceptable plan has been approved and incorporated into the terms of award.”64

60 http://www.ijdc.net/index.php/ijdc/article/view/8.2.194/327 61 http://www.asis.org/Bulletin/Aug-14/AugSep14_Kozlowski.html 62 http://www.esrc.ac.uk/_images/Research_Data_Policy_2010_tcm8-4595.pdf 63 http://www.bbsrc.ac.uk/web/FILES/Policies/data-sharing-policy.pdf 64 http://www.genomecanada.ca/medias/PDF/EN/DataReleaseandResourceSharingPolicy.pdf

24

Other funders have chosen not to include the DMP as a part of the proposal evaluation at all. The NIH states that reviewers will not factor the proposed data-sharing plan into the determination of scientific merit or priority score. However, program staff members are responsible for overseeing the data-sharing policy and assessing the appropriateness of the plan. In other words, NIH staff will evaluate the content of the plan based on whether it provides comprehensive information about how researchers will manage the data. Any concerns must be resolved prior to making any award.65 Presumably, the NIH has developed the expertise internally taking the responsibility off of the peer-review committees.

In general there are four potential options for evaluating DMPs in the proposal stage:

1) The DMP is reviewed as part of the excellence review. Assessment is a full-weight component of the excellence assessment and can impact the adjudication

2) The DMP is reviewed separately from the excellence review, but with an impact on acceptance of proposal

3) The DMP is reviewed separately from the excellence review and has no impact on acceptance of proposal

4) No review process

Table 5 highlights the approaches of the different funders where information is available.

Table 5: Funders approaches to assessing DMPs

Funder Assessed as part of peer review process

Assessed separately, impact on proposal

Assessed separately, no impact on proposal

Not assessed during peer review process

Genome Canada X

EC- Horizon 2020 X

UK- BBSRC X

UK- CRUK X

UK- ESRC X

UK- MRC X

UK- STFC X

US- NIH X

US- NSF X

It can be challenging to assess data management plans as part of a funder’s peer review process, depending on the discipline and the expertise of committee members. In the scenario where the peer review committees are reviewing the plans, they may not have sufficient knowledge to be able to determine quality. It has been reported, for example, 65 http://grants.nih.gov/grants/policy/data_sharing/data_sharing_faqs.htm#901

25

that reviewers of NSF proposals (from disciplinary peer-review committees) who rely on the general policy guidelines have found it difficult to identify the components of a good data management plan.66 To address this, some organizations have developed guidance for peer reviewers. John Hopkins University Library developed a checklist to assist NSF proposal reviewers67 and both the MRC and ESRC in the UK have published guidance documents to assist their peer review committees in evaluating the quality of DMPs.

5.3 Monitoring Compliance A variety of approaches are also being used to monitor adherence to the policies. Most commonly, researchers/projects are required to provide a written report about how they have adhered to the policy requirements in their final reports. According to the DCC, several of the UK funding councils are actively monitoring compliance with their policies via the final report process. The Engineering and Physical Sciences Research Council (EPSRC) is monitoring progress and compliance on a “case-by-case basis”. Both the Economic and Social Research Council (ESRC) and the Natural Environment Research Council (NERC) state that they are prepared to withhold the final grant payment if data are not properly managed and offered for deposit. However, the extent to which such penalties are applied is unclear.”68 In addition, ESRC expects grant holders to report about “the on-going implementation of the data management and sharing plan through annual reporting to ESRC”.69 It is unclear to what degree and through which methods the US funders and other agencies are monitoring compliance.

Ultimately, comprehensive monitoring will require that data sets can be tracked. There are a number of tools that are emerging to help improve the discoverability of datasets. Data citation using permanent identifiers, such as Digital Object Identifiers (DOIs) provides a permanent ID for datasets, which is helpful for tracking data that has been deposited into a public repository. In addition, there are initiatives in both Australia and the UK to improve the visibility and discoverability of data sets, both to support access and re-use. ANDS has developed a dataset registry, called Australian Research Data Commons, to make better use of Australia's research data outputs.70 Similarly, Jisc (a membership organization that supports the use of digital technologies in UK education and research community) and the DCC are currently working on a UK registry, which will “provide a coherent point of access to discoverable, searchable, browsable and actionable descriptions of given datasets and how to access them, and so showcase the wealth of UK research data”.71 These registries not only assist policy makers in tracking datasets, but also contribute to the ultimate aims of RDM policies by improving the discoverability of research data and supporting re-use. They also contribute to a system where data can be cited.

66 http://www.nsf.gov/pubs/policydocs/pappguide/nsf13001/aag_6.jsp#VID4 67 http://dmp.data.jhu.edu/resources/grant-reviewers-guide/ 68 http://www.dcc.ac.uk/sites/default/files/documents/RC%20policy%20overview%20v2.2.pdf 69 http://www.esrc.ac.uk/_images/Research_Data_Policy_2010_tcm8-4595.pdf 70 http://researchdata.ands.org.au 71 http://www.dcc.ac.uk/projects/research-data-registry-pilot

26

5.4 Confidentiality and Intellectual Property Almost all RDM policies contain clauses addressing the issues of privacy and intellectual property, and some also have clauses dealing with other types of sensitive data. Referred to as “ethical open access” or “intelligent open access”, policies must aim to strike a balance between the rights and interests of investigators, study participants, and the public. This balance is not always easy to achieve.

In Canada, university research ethics boards are very concerned with ensuring privacy of study participants and often enforce stringent practices restrict researchers’ abilities to share data. They base their approach on the “Tri-Council Policy Statement on the Ethical Conduct for Research Involving Humans” (TCPS), which sets out privacy and confidentiality requirements for researchers working with human participants, including for secondary use of research data. The policy statement emphasizes that respect for privacy in research is an internationally recognized norm and ethical standard. These codes can conflict with data sharing policies, if applied to stringently. For example, research ethics boards may request a plan for data disposal, which could conflict with a funding agency policy that requires data retention and sharing.

There are, however, established best practices to ensure that the confidentiality of study participants is protected. In cases where the data cannot be modified to protect confidentiality without significantly compromising the research potential of the data, data is restricted and confidentiality safeguards are imposed. Many funding agencies therefore require practices such as anonymization to be adopted by researchers before releasing data. In cases where anonymization is not possible, then researchers must explain why in their DMP or grant application. For example, the Wellcome Trust deals with confidentiality in the following way:

“In designing studies, researchers must ensure that they have appropriate systems to protect the confidentiality and security of data pertaining to human subjects, and minimise any risks of identification by data users. This can be achieved through the use of appropriate anonymisation procedures and managed access processes. Such systems should be sufficient to safeguard participants, but proportionate to the level of sensitivity of the data and associated risk. They should not unduly inhibit responsible data sharing for legitimate research uses.”72

In terms of intellectual property, many policies have exceptions for data that have potential commercial value. In these cases, they try to strike a balance between the value of broad data sharing and deriving any commercial benefits from the research. Approaches to protecting IP usually involve implementing embargo periods to allow for patent applications. For example, the NSF policy states:

“It is NSF’s strong expectation that investigators will share with other researchers, at no more than incremental cost and within a reasonable time, the primary data, samples, physical collections and other supporting materials created or gathered in the course of work under NSF grants. However, it is also necessary to protect intellectual property rights and potential commercial value. The Data Management Plan should describe the proposed approach, which will then be subject to peer review and program management. (For example, research use of sensitive data is

72 http://www.wellcome.ac.uk/About-us/Policy/Spotlight-issues/Data-sharing/Guidance-for-researchers/index.htm#five

27

often allowed through reasonable binding agreements that contain confidentiality provisions.)”73

The NIH “recognizes that the investigators who collected the data have a legitimate interest in benefiting from their investment of time and effort. NIH continues to expect that the initial investigators may benefit from first and continuing use but not from prolonged exclusive use.”74

In the UK, IP related to research falls under the auspices of the institution. Therefore, the UK funding councils tend to rely on institutional policies to set requirements for IP issues around data. In Canada, at most universities it is the researcher/creator who owns the IP, although there are some exceptions to this.75

A few policies also include exceptions for “sensitive” data. This is particularly important in research areas dealing with indigenous knowledge and national security. “The Statement of Principles and Practices for Arctic Data Management” published in April 2013 by the International Arctic Science Committee (IASC) provides a framework for handling data collected from indigenous communities, (as does guidance contained in the TCPS2):

“In the context of research involving Indigenous knowledge, data management principles based on the concepts of respect, reciprocity, and responsibility should be observed. This includes appropriate engagement of Indigenous people, communities or organizations throughout the entire data life cycle, formal attribution of contributed knowledge, establishment of informed consent for use of knowledge and derived products, and the maintenance of contributor control of data and information resources. Required institutional ethics review processes (e.g. Institutional Review Boards, Research Ethics Boards etc.) will guide data management, however Indigenous communities or organizations may have specific practices or requirements in place. It is the responsibility of researchers to familiarize themselves with and adhere to these practices and requirements.” 76

5.5 Multiple Policies Co-funding is becoming a common practice and many researchers may find themselves in a position where they are subject to the requirements of two or more funding agencies, in addition to the requirements or their institution and journal. This can be a particularly problematic issue for researchers that are co-funded by government and private industry, since companies may seeks to protect intellectual property resulting from research, which could include the data. In these cases, researchers could be asked to create interagency agreements concerning data management that would be shared and approved by all funders.

A few funders have developed guidance for researchers about how to adhere to policies in the case of multiple funders with differing requirements. The NIH advises grantees as

73 http://www.nsf.gov/bfa/dias/policy/dmpfaqs.jsp 74 http://grants.nih.gov/grants/policy/data_sharing/data_sharing_guidance.htm#time 75 http://blogs.sfu.ca/departments/cprost/wp-content/uploads/2012/10/IP-Policy-Introduction-January-2010FINALCombined.pdf 76http://www.innovation.ca/sites/default/files/Rome2013/files/IASC%20Statement%20on%20Arctic%20Data%20Management_2013.pdf

28

follows, “[t]he NIH recognizes that there may be circumstances where a co-funder has requested restrictions on data sharing as a condition of funding. These restrictions should be identified in the application and a proposal made about how data from the co-funded project will be shared. Should you believe that you are unable to share any of the data, your justification will be considered by NIH program staff.”77

Harmonization of approaches and policies across agencies (both in Canada and internationally) can go a long way to help address some of these issues that arise when dealing with multiple funders.

5.6 Costs Adhering to research data management policies can incur extra costs. A crucial factor is the type of data created. The larger and more complex the data being managed, the greater the effort required and the greater the potential costs.78 Most funders reviewed in this scan consider data management activities as being eligible expenses within a project budget.

In addition, there are funders that fund and maintain data repositories to support data archiving, although the scope of these domain repositories do not cover all datasets produced through their funded research.

There are costs for data management that fall across the entire data lifecycle. The UK Data Service has compiled a detailed list of the costs associated with research data management that are incurred during the life of the project. 79

Table 6: Costs for Research Data Management

Data storage

Data transfer and access

Data backup

Data security

Consent for data sharing

Transcription

Anonymization

Data sharing

Operationalizing data

Data description

Data cleaning

Data documentation

Metadata

Formatting and organizing

Digitization

Data format

The real costs for any given project will depend to a large extent on the nature of the data collected. More information about potential costs is provided in the Implementation Challenges section of this report.

77 http://www.data-archive.ac.uk/media/247429/costingtool.pdf 78 http://www.ed.ac.uk/schools-departments/information-services/research-support/data-management/how-manage-data 79 http://www.data-archive.ac.uk/media/247429/costingtool.pdf  

29

6. Approaches to Policy Implementation This section describes the various approaches taken to implementing RDM policies in different jurisdictions. Regardless of approach, experiences of others demonstrate that full adherence to policies takes time.

6.1 Engineering and Physical Sciences Research Council (UK) In the UK, the funding agencies have taken a strong stance through comprehensive policy adoption across all agencies based on a set of common principles. In 2011, the seven RCUK councils adopted a set of common principles for research data management. Each council was expected to develop a policy that adheres to those principles within a certain time frame.

The policies across the 7 agencies differ and reflect the specific disciplinary context served by each council. 6 of the councils have requirements for data management plans, expect research projects to manage their data according to certain standards, and require them to share data at the end of a project. The exception is the EPSRC, which has taken a different approach by placing some of the responsibility for adherence to the policy on the institutions. In April 2011, all UK Vice Chancellors received a letter from the EPSRC, that sets out 9 “expectations” of organisations in receipt of EPSRC research funding which are summarized below: 80

1. Research organisations will promote internal awareness of these principles and policies

2. Published research papers should include a short statement describing how and on what terms any supporting research data may be accessed

3. All of their researchers or research students funded by EPSRC will be required to comply with research organization policies in this area or, in exceptional circumstances, to provide justification of why this is not possible.

4. Publicly-funded research data that is not generated in digital format will be stored in a manner that facilitates it being shared should a valid request for access to the data be received

5. Research organisations will ensure that appropriately structured metadata describing the research data they hold is published (normally within 12 months of the data being generated) and made freely accessible on the Internet. Where the research data referred to in the metadata is a digital object it is expected that the metadata will include use of a robust digital object identifier (e.g. DOI from DataCite)

6. Where access to the data is restricted the published metadata should also give the reason and summarise the conditions

7. Research organisations will ensure that EPSRC-funded research data is securely preserved for a minimum of 10 years

8. Research organisations will ensure that effective data curation is provided throughout the full data lifecycle

9. Research organisations will ensure adequate resources are provided to support the curation of publicly-funded research data

80 http://www.epsrc.ac.uk/about/standards/researchdata/expectations/

30

The framework also requires organizations (rather than researchers) to identify and map out the steps they will take to achieve full compliance to the Roadmap. The EPSRC has given the universities a deadline of May 1st, 2015 to achieve full compliance.

The EPSRC has stated that it may request to see individual roadmaps on a case-by-case basis and could require evidence of activity to achieve compliance at any time. If, after May 2015, an institution is found to be deliberately obstructing the sharing of research data or otherwise seriously failing to comply with the ESPRC’s expectations, then the EPSRC may ultimately withdraw its funding.81

This framework has been the impetus for a number of UK universities to invest in research data management, including services, infrastructure and storage and the DCC now lists 10 institutional roadmaps that have been made public with more in development.

According to a report published by the Society for Research into Higher Education in March 2013,

“This move on the part of the EPSRC […] is pushing universities to review their research data management practices, develop Research Data Management policies, and investigate the resource and infrastructure implications. Some universities have already introduced data preservation and sharing policies requiring their researchers to address, at the outset of their projects, the question of data management and sharing (e.g. Universities of Edinburgh and Oxford); and some have developed institutional data repositories in which academics and PhD students are encouraged to deposit their data (e.g. ‘Edinburgh DataShare’). One of the ways in which these new requirements are being institutionalised by universities is by defining data sharing as ‘good research practice’ and incorporating data management and sharing into university ethics and research governance regulations and procedures.”82

6.2 Pilot on Open Data (European Commission) The European Commission (EC) has chosen to introduce an RDM policy incrementally beginning with a pilot project, rather than implementing across-the-board requirements for all EC-funded research projects. The “Pilot on Open Research Data” targets specific research areas (listed below):83

• Future and Emerging Technologies • Research infrastructures – part e-Infrastructures • Leadership in enabling and industrial technologies – Information and

Communication Technologies • Societal Challenge: Secure, Clean and Efficient Energy – part Smart cities and

communities • Societal Challenge: Climate Action, Environment, Resource Efficiency and Raw

materials – with the exception of topics in the area of raw materials • Societal Challenge: Europe in a changing world – inclusive, innovative and

reflective Societies • Science with and for Society

81 http://www.bath.ac.uk/rdso/assets/pdf/University-of-Bath-Roadmap-for-EPSRC.pdf 82 http://www.srhe.ac.uk/downloads/MauthnerScopingReport.pdf 83 http://europa.eu/rapid/press-release_IP-13-1257_en.htm  

31

The pilot areas correspond to about €3 billion or 20% of the overall Horizon 2020 budget in 2014 and 2015.

The aim of the Pilot is to give the Commission a better understanding of what supporting infrastructure is needed and of the impact of limiting factors such as security, privacy or data protection or other reasons for projects opting out of sharing. It will also contribute insights into how best to create incentives for researchers to manage and share their data. The pilot will be monitored throughout Horizon 2020 with a view to developing future policy and EU research funding programs.

6.3 Code of Conduct and the Australian National Data Service In Australia, the “Australian Code for Responsible Conduct of Research”84 places the onus of responsibility on the universities for managing and preserving data. As discussed earlier, the code requires institutions to retain research data, provide secure data storage, identify ownership, and ensure security and confidentiality of research data. Although the Code is not applied in a strict manner, it has been an inducement for Australian universities to develop RDM services, resulting in more robust repositories and services at Australian institutions than in many other jurisdictions.

Also contributing to a more robust RDM environment is the fact that the Australian government has invested significantly in data management via the Australian National Data Service (ANDS).85 Research Data Australia is the flagship service of ANDS, which provides a comprehensive window into the Australian Research Data Commons enabling Internet-based discovery to Australia’s data, projects, researchers and institutions.86 Currently, the service aggregates metadata from over 90 collections across Australia, including 22 universities and numerous domain research data centres. ANDS also invests in the development of domain and institutional repositories as well as training and support for RDM. 6.4 Office of Science and Technology Policy (United States) In February 2013, The White House’s Office of Science and Technology Policy (OSTP) released a policy memorandum that directed all 22 federal agencies with more than $100M in R&D expenditures (including the NSF and NIH) to develop plans to make the published results of federally funded research freely available to the public within one year of publication and requiring researchers to better account for and manage the digital data resulting from federally funded scientific research.

In a March 2014 letter to the House and Senate Appropriations Committees, the director of the Office of Science and Technology Policy (OSTP) reported that “all agencies subject to the requirements in the memorandum have now submitted draft plans” and “are currently revising their plans to address OSTP and OMB [Office of Management and Budget] comments and ensure compliance with all of the requirements laid out in the OSTP memorandum.”87

84 http://www.nhmrc.gov.au/_files_nhmrc/publications/attachments/r39.pdf 85 http://ands.org.au/resource/ands-business-plan-2013-14.pdf 86 http://researchdata.ands.org.au/home/about 87 http://www.icpsr.umich.edu/icpsrweb/content/datamanagement/ostp.html

32

Although the agencies are working together to try to align their policies in terms of data management, it is likely that there will be significant variations in plans across these agencies, making for a confusing environment for researchers who will be required to adhere to policies.

Little to no extra support and funding has been provided to support the implementation of policies.

7. Implementation Challenges A 2013 survey of over 300 Canadian researchers from across disciplines undertaken by Susan Mowers et. al. provides some indication that many researchers do not use repositories to share their data. The survey found only 4% of respondents shared their data through a “curated digital data repository” 88, and another 14% used an institution repository or a “public domain archive” 81% of respondents indicated that they stored data on their local hard drives.89

7.1 Disciplinary Contexts Implementation challenges must be viewed through a disciplinary lens. Across domains, disciplines and sub-disciplines, the types of data produced and used are extremely diverse; standards differ significantly, as does the availability of infrastructure. In some fields researchers already have a well-established culture of data sharing, there are well-established practices, and support and infrastructures to allow data sharing. In other fields no such mechanisms exist, and others fall in between these two extremes. Many of the challenges with implementing RDM must also be viewed through the particular disciplinary context. “Barriers to effective data sharing and preservation are deeply rooted in the practices and culture of the research process as well as the researchers themselves.”90

7.2 Researcher Preparedness

Researchers’ perspectives towards data sharing are very discipline specific. Surveys and interviews undertaken over the last decade have articulated a wide range of opinions on the topic which cannot be easily generalized into a single statement about researchers’ attitudes. Typical objections to data sharing include data ownership and fears of being scooped; the time and skills involved with managing data; and issues of privacy involving data about human participants. A review of the literature across 15 international jurisdictions undertaken in the Netherlands found that “although there are major differences in the way disciplines conduct their research, they also have a number of factors in common when it comes to data storage and access. They all encounter both

88 A curated archive refers to a repository where there is active management and appraisal of data over the lifecycle of scholarly and scientific materials - http://digitalcuration.blogspot.ca/2009/08/curated-databases-and-data-curation.html 89 http://gsg.uottawa.ca/data/open/aa-interim-survey-report/20130801-en.pdf 90  Tenopir (2011) et. al. http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0021101  

33

technical barriers, for example the use of obsolete software, and non-technical ones, such as fear of competition, lack of trust, lack of incentives, and lack of control.”91

Expertise in the research community is also an important barrier. A survey in the US of researchers at five different institutions found that none of the researchers interviewed had received formal training in data management practices. “None of the scholars interviewed during this study expressed satisfaction with their level of expertise in data management, and few had access to individuals who could provide knowledgeable guidance. On the contrary, most participants reported feeling adrift when establishing protocols for managing their data and added that they lacked the resources to determine best practices, let alone to implement them.”92

Many disciplines still lack formalized and standardized procedures for managing research data. There are also large gaps in terms of training for data management. According to a report published by Knowledge Exchange, a multi-national, European co-operative effort that supports the use and development of information and communications technologies (ICT) infrastructure for higher education and research, the most important challenge to data sharing is that it is not yet very common among scholars and is not yet seen as a regular activity among scientists.93 In interviews they conducted, the main hurdle in data sharing is the individual scientist who is reluctant to put effort into data sharing. “This is mainly for cultural reasons: ownership of the data, workload to properly curate the data making them available for others, and lack of career-reward for making this type of effort.”94 Ongoing promotion and education of researchers will be needed to address these barriers.

7.3 Incentives Equally significant is the lack of incentives and rewards for data management and sharing. Incentives would greatly accelerate the adoption of RDM practices in the research community and these rewards should be part of the formal evaluation processes at funding agencies and institutions. While far from common, we are beginning to see this happen in some contexts. In 2013, for example, the NSF began to allow datasets to be cited as relevant work products in biographical sketches independent of the related publication(s).95 The journals that require data sharing can also be an important incentive for researchers, as they ask that researchers deposit their data in order to have their articles published. Services that offer a permanent URL for a dataset, such as DataCite’s DOIs, enable citations that can then be used to track and acknowledge the re-use of data. These mechanisms are “paving the way for new metrics and publication models that recognize and reward data sharing” but without actually developing any of these indicators. “The availability of proper metrics can help researchers to make their data work more visible. This may subsequently act as an incentive for more data sharing and in this way a virtuous circle may be set in motion.”96

91 SURF Foundation: What Researchers Want. www.surffoundation.nl/en/publicaties/Pages/Whatresearcherswant.aspx 92 http://www.clir.org/pubs/reports/pub154/problem-of-data 93 Ibid 94 Ibid 95 http://www.asis.org/Bulletin/Aug-14/AugSep14_Kozlowski.html 96 http://www.knowledge-exchange.info/datametrics

34

7.4 Costs

Costs for data management are often divided into two aspects: the costs of managing and preparing data during the research project; and the costs of providing access and preservation to data once the project is over.

In terms of project-based data management activities, the UK Data Service has developed a costing tool to help researchers anticipate costs of RDM97. The tool identifies costs across the data lifecycle, which will be unique to each project depending on the size and complexity of the datasets collected. The costing tool was developed with input from researchers, who were asked to estimate the time or cost needed for activities related to: data collection, data entry and transcription, data validation and documentation and the cost of preparing data for archiving and re-use. The researchers participating in the process “found it hard to cost data management activities, as many activities are an integral part of standard research activities and data analysis.”

While the costs of managing data will differ depending on volume and complexity of data, one general rule of thumb suggested via the Jisc Research Data Management listserve is that 5% of the project costs will be for data management activities in cases where “data have high re-use potential, and the data have at least some features (anonymization, complex documentation, size) that might make data preparation more costly.” 98 In other cases, where data is not subject to such complicated preparation, the costs could be significantly lower. These costs reflect the costs of data management during the lifespan of the project.

There is a second part of the cost scenario, which represents the costs of preserving and providing access to data after the project is over. These generally fall under the costs of maintaining a data repository. These costs can also vary significantly depending on what level of curatorial service is attached to the repository. In 2012, the Royal Society in the UK undertook an analysis of the costs involved in the long-term management of data.99 They categorized 4 tiers of data management: Tier 1 and 2 are represented by major international data initiatives that have well defined protocols for the selection and incorporation of new data and access to them (e.g. genomics data) and data centres and resources managed by national bodies (such as UK Research Councils or prominent research funders such as the Wellcome Trust). Tier 3 is curation at the level of individual universities and research institutes, or groupings of them; and Tier 4 is when the individual researcher or research group collates and stores its own data, often making it available via a website to collaborators or for public access. The analysis found that the costs of running a curated data archive100 ranged from $350,000 US per year for the Dryad repository to $6-7 million for the Worldwide Protein Data Bank. In terms of institutional data repositories, costs were generally lower, but real costs were difficult to determine because institutional data repositories often share infrastructure and staff cross other positions in the same institution.

97 Ibid 98 RESEARCH-DATAMAN@jiscmail.ac.uk discussion list" <RESEARCH-DATAMAN@JISCMAIL.AC.UK, October 8, 2014 99 https://royalsociety.org/policy/projects/science-public-enterprise/digital-repositories/ 100 A curated archive refers to one where there is active management and appraisal of data over the life-cycle of scholarly and scientific materials (http://digitalcuration.blogspot.ca/2009/08/curated-databases-and-data-curation.html)

35

7.5 Institutional Role Universities have not traditionally seen research data management as part of their mandate and have been relatively slow to become engaged. This is beginning to change. There has been gradual growth in RDM services provided by universities, usually through the libraries, and it is likely that this trend will continue and expand. In the UK and Australia, many of the larger institutions have already implemented fairly robust services. Funding agency requirements that place some responsibility for data management on the institution have been the impetus for adoption of RDM services. In the US and Canada, several universities are providing support services for researchers, and a few are managing data repositories.

Universities are important stakeholders in the area of research data management activities. They have direct access to researchers (unlike funders) and can raise awareness of the benefits of RDM, provide support and guidance, and collect data in local repositories. A 2012 UK report argued “[u]niversities and research institutes should play a major role in supporting an open data culture by [...] developing a data strategy and their own capacity to curate their own knowledge resources and support the data needs of researchers.”101

In 2013, the German Rectors Conference, an association of 268 universities in Germany issued a resolution102 about research data management that urged universities to:

• Agree on guidelines about how to handle digital research data • Collaborate beyond the boundaries of the university • Improve information skills • Develop institutional infrastructures for research data management

The role of the institution was also underscored in the 2011 report of Canada’s Research Data Summit, which stated that institutions should:

• Maintain sustainable research data repositories • Support the implementation and enforcement of funding agency data policies • Provide support on campus for data management activities through employment

of trained data scientists • Implement rewards for data management and include these in promotion and

tenure processes

One of the reasons that institutions have been reluctant to get involved in research data management is that they must find and justify resources to devote to this area. Most Canadian institutions that are providing RDM services have redirected some of their budget from other services, rather than obtaining new monies from other funding streams, although a few have received funding via other sources.

101 http://royalsociety.org/policy/projects/science-publicenterprise/report/ 102 From a presentation by Jochen Schirrwagen at RDA Long Tail of Research Data, Amsterdam, September 23, 2014  

36

8. Current State of RDM in Canada In Canada, over the past fifteen years, there have been numerous consultations and meetings discussing the state of research data management in the country and proposing various solutions. The more recent events, summarized from the comprehensive account written by Chuck Humphrey 103 , include the National Data Archive Consultation in 2002, which produced a report calling for the adoption of a national data archive service to collect and preserve the research data produced in Canada.104 In 2004, there was a National Consultation on Access to Scientific Research Data (NCASRD), which aimed to address the issues of data access in the physical and life sciences. The report105 called for the establishment of a national steering body to help coordinate data management and preservation services. In 2008, the Research Data Strategy Working Group was launched to bring together the major stakeholder communities and develop strategies for improving the situation of RDM in Canada. They hosted the National Data Summit in 2011, which brought together over 160 senior managers. The final report published by the group, “Mapping the Data Landscape: Report of the 2011 Canadian Research Data Summit”, included a set of recommendations to develop stronger community involvement in research data management and preservation. This led to the launch of Research Data Canada, an organization that is working to move forward on the recommendations in the report. And, in 2014, the Digital Infrastructure Leadership Council hosted a national meeting to discuss better coordination around Canada’s ecosystem, including the management of research data.

In 2013 the TC3+ (CIHR, NSERC, SSHRC, CFI) and Genome Canada published a document that proposed “changes to their funding policy frameworks that promote excellence in data management practices, thereby advancing digital scholarship and Canada’s digital infrastructure ecosystem to the benefit of Canadians”. This document identified three important next steps for the members of TC3+:

1. Define the core elements of an agency-based and focused data stewardship plan

2. Work with other organizations and working groups to ensure ongoing consultation and coordination with all stakeholders, including the provinces, in the development of Canada’s national digital infrastructure for research

3. Collaborate in the development of a coordinated plan to encourage the establishment of new and/or the enhancement and sustained operation of existing world-class centres specializing in data management

This was followed by a consultation with the stakeholder community. While none of these initiatives have resulted in an immediate or profound change in the research data management environment yet, they have contributed to a slow but steady increase in the visibility of research data management as an issue.

8.1 Gap Analysis RDM policies cannot be adopted in isolation. Good research data management practices depend on multiple factors being in place – including incentives, skills and expertise, services, infrastructure, funding, and policies. These factors create a setting that

103 http://preservingresearchdataincanada.net/category/introduction/ 104 http://www.sshrc-crsh.gc.ca/about-au_sujet/publications/da_phase1_e.pdf 105 https://datalib.library.ualberta.ca/data/NCASRDReport_e.pdf  

37

supports RDM across the lifecycle and research domains. This gap analysis reviews the current situation in Canada across four axes representing the key factors that will contribute to the successful implementation of an RDM policy:

• Funding for RDM across the data lifecycle • Infrastructure and services for RDM • Expertise and support for the proper management of data • A shared understanding of roles and responsibilities of the different players

Funding: Funding in Canada for data management activities is not consistently available across the lifecycle and disciplines. As discussed earlier in the document, the costs of managing research data can be significant depending on the types of data produced. Costing tools developed in Australia, UK and the US illustrate a variety of costs across the entire data management lifecycle including production, dissemination, sharing, and preservation. Most funding for research data management in Canada is available during the lifespan of the project. For example, the costs of data management in the collection and analysis phase of the research project are generally considered eligible expenses in most grant programs, as are the costs for the development of databases and the storage of that data. However, once the project is over, data must be archived and handed over to a long-term data repository that provides access and preservation services in order to enable its further re-use.

While there are many possible models for funding RDM services and infrastructures, Canada currently has only a few select mechanisms to support data management beyond the lifespan of the project. There is funding for some large domain data centres through CFI and direct government funding, as well as the indirect costs of research, however, these do not cover all domains nor, in some cases, do they support long term access. The lack of funding models in Canada was identified as an important issue at the Digital Infrastructure Summit held in January 2014. The summit’s report states, “there are few vehicles of support for the system-wide elements of RDM. Further, some small incremental funding was needed for certain aspects of the design and implementation of DI [Digital Infrastructure]. Although the latter amounts are not likely to be significant, they are not currently “line items” in the budget of any of the key organizations. This gap needs to be addressed.”106

Infrastructure and services: The infrastructure for research data management in Canada is piecemeal, with some fields having very good coverage and others very little. Domains that are well covered are generally those that have access to large national repositories and have established traditions of data sharing (e.g. astronomy, ocean science, Statistics Canada Data Centres, polar/arctic research data, genomics). There are also large-scale international data repositories that preserve and provide access to data in their fields (e.g. PubChem, GenBank, Protein Data Bank, Global Biodiversity Information Facility, Inter-university Consortium for Political and Social Research). The Canadian government maintains repositories that house data in many areas deemed of national importance. In terms of multidisciplinary repositories, both Figshare and Dryad, which offer more generic repository services, are available to Canadian researchers to deposit and several

106 http://digitalleadership.ca/wp-content/uploads/2014/02/Summary-Report-of-Summit-2014-Final-March-2014.pdf

38

institutions are running repository services that collect and provide access to the data produced by researchers on their campus (University of Alberta, Simon Fraser University, Scholars Portal in Ontario, University of British Columbia, University of PEI).

RDM can be very complex and support services may be required at various stages throughout the data lifecycle, from preparation of data management plans, to documentation of data for access and preservation, to the re-use and analysis of datasets. There is some support in the context of domain repositories, as well as a growing number of libraries that are offering support to the researchers at their institution, but there remain numerous gaps in the infrastructure and services required to comprehensively support RDM across all communities.

A recent initiative, now referred to as Portage (formerly Project ARC), is developing a national library-based network for research data management in Canada to address both the infrastructure and services gap. 107 Portage is managed by the Canadian Association of Research Libraries and has two components: a national network of expertise and a national preservation and discovery system. The network is in its early stages, but the aim is to provide support for data management planning across Canada, as well as begin to build the infrastructure that would support widespread data management, sharing and preservation. Portage is working closely with several universities, Research Data Canada and Compute Canada to develop solutions that will improve our ability to store and preserve the range of research data produced in Canada.

Skills and expertise: Research data management requires specialized skills and knowledge, both for the researchers handling the data and for the support services (such as libraries and IT staff). RDM expertise must be embedded throughout the lifecycle of the research data, from their collection, to dissemination, to preservation. In a blog post about RDM, Chuck Humphrey, Data Librarian at the University of Alberta and one of Canada’s experts on research data management, states, “[d]ata management activities span the research lifecycle and involve many different skills, drawing upon a variety of expertise. The demands for data management expertise depend on the scale of the research project. A small project may involve only a couple of people, who can manage with a general set of skills. A much larger project may require a team of experts with each team member responsible for a specific specialization.”108

There are some efforts to improve data management support for researchers. Many Canadian universities are developing services to provide data management support on campus, which range from basic information resources to more comprehensive support, such as provided by the University of Alberta, which offers expert advice, repository services, and data management planning tools.109 Portage is an initiative being led by the Canadian Association of Research Libraries which aims to launch a distributed national network of expertise that would offer information resources, training and consulting services for Canadian institutions, libraries, and researchers. In addition, CARL has offered a course for librarians on RDM services and is organizing another course about research data management planning in 2015.

Despite these efforts, there is still much work to be done to improve awareness and expertise for RDM in Canada. The 2011 Gap Analysis observed “[r]esearchers rarely have the skills to appropriately manage their data and there are few data professionals 107 http://data-carl-abrc.ca/project-arc/  108 http://preservingresearchdataincanada.net/2012/12/17/research-data-management-infrastructure-iii/ 109 http://www.library.ualberta.ca/researchdata/

39

to assist them.”110 In a report published in January 2013, the Research Data Canada Education and Training Committee went on to say, “[c]urrently within Canada there are few training opportunities available in the area of RDM. Canada lacks the national-level coordination of a body like Jisc in the UK, and federal granting councils do not yet provide the necessary policy incentives regarding RDM.”111 In their report, the RDC Committee made a number of recommendations about how Canada can improve the current situation, including, “Canada needs a multipronged strategy for the delivery of education in order to build capacity for research data stewardship in Canada. This should include integrating RDM in graduate curricula for future researchers, implementing RDM courses in information schools and other relevant academic programs, and providing a variety of training that will assist current researchers, librarians and other stakeholders to up-skill for RDM.”112

Roles and responsibilities: Data management is rarely the sole responsibility of the principle investigator. A number of different stakeholders are involved in the research process and have a role to play in ensuring good practices. Other stakeholders include institutional leaders; co-investigators and graduate students; external contractors involved in data collection; research administrators; institutional IT services; and institutional or external data repositories. Table 6 is adapted from a list of responsibilities published in 2007 by UKOLN113. It outlines the responsibilities of the key players involved with research data management and identifies key obstacles for taking on responsibilities.

Table 7: Roles and responsibilities in RDM

Role Responsibilities Obstacles

Researcher Manage data for the life of the project

Meet standards for good practice

Comply with funder and institutional data policies

Work up data for use by others

Low awareness of appropriate standards in some fields

Low knowledge about good data management practices in some fields

Lack of funding for support by data scientists

Lack of time to properly document data

Universities Adopt data management policies

Raise awareness of funder requirements

Ensure standards of good practice are met

Provide training and advice to researchers

Manage a repository service for long term access and preservation of data

Only gradual uptake of RDM policies and services at institutions (including data repositories)

Training opportunities for RDM support is not widespread

110 http://www.carl-abrc.ca/uploads/pdfs/data_brochure-e.pdf 111 Summit Report 112 Ibid 113 http://www.ukoln.ac.uk/ukoln/staff/e.j.lyon/reports/dealing_with_data_report-final.pdf  

40

Data centre/ repository

Manage data for the long-term

Meet standards for good practice

Provide training for deposit

Promote the repository service

Protect rights of data contributors

Provide tools for re-use of data

Lack of sustainable funding for data centres in many fields

Funder Adopt data management policies

Monitor and enforce data policies

Fund data management activities as part of the project

Resource post-project long-term data management

Support workforce capacity development of data curators

Unwillingness to divert funds from research to data management

Currently, the roles and responsibilities outlined in the table are aspirational. There is no common understanding across stakeholders about where the responsibilities lie for the various aspects of RDM. Both researchers and institutions are apprehensive about taking on greater responsibility for managing research data. Researchers are worried about the time and resources required for preparing data, as well as a pervasive lack of expertise within the research community. Institutions are concerned about how they will fund data management support services and repositories.

In terms of coordination across stakeholder communities, Research Data Canada has been active in bringing together different stakeholders to move towards a common understanding of roles and to develop collaborative solutions.

8.2 Readiness for Policy Implementation In the previous section of the report, we discuss the current state of RDM across a number of indicators. That analysis informs the assessment of national readiness for RDM policy implementation in Canada. There are a number of ways in which readiness can be considered. For a detailed assessment, one could use the Community Capability Model Framework. This framework identifies eight capabilities by which capacity for RDM can be measured: Openness, Legal, Ethical and Commercial Considerations, Collaboration, Economic and Business, Skills and Training, Common Practices, Research Culture and Technical Infrastructure.114 However, this kind of assessment involves a very detailed, intensive process that goes beyond the scope of the work for this brief. The following table provides a preliminary assessment of the community’s readiness according to key policy elements. It should be noted that given the large variations in disciplinary practices and infrastructure, any assessment will have limitations and offers a somewhat general view about readiness to adhere to policy requirements.

114 http://ozk.unizd.hr/proceedings/index.php/lida/article/viewFile/121/123

41

Table 8: Readiness assessment scores

Policy element

Readiness assessment

Data quality and standards

Standards for the collection of research data vary significantly across the disciplines. Some fields already have long established standards while other fields are still developing best practices.

International surveys have found that many researchers do not feel they have the expertise and knowledge to appropriately manage their data. Therefore, support and training for RDM in the research community will be required.

Some Canadian universities already provide services through their libraries to provide guidance for researchers in managing their data. For example, a review undertaken by the Research Data Canada Standards and Interoperability Committee found that at least 21 universities are providing some resources about data management planning.115 Portage is developing a model for a network of expertise that will build on and expand the scope of existing university services and offer information resources as well as in-depth support for researchers across the country from experts in the library community.

In some cases, applying appropriate standards and quality control may require extra funding for data management activities. These activities are not always eligible expenses in the grant.

In general, researchers can be expected to identify and use standards and best practices for managing research data in their field if support is for this is available.

Data access and sharing

There are significant gaps in the repository infrastructure in Canada. These gaps are being addressed through Portage and other repository initiatives, but it will take time and funding to build comprehensive repository network across the country.

Another challenge is overcoming the reluctance of some researchers to share their data. There are a number of issues that contribute to this situation, including lack of policies and incentives, the time and effort it takes to prepare the data, and the fear of being scooped.

The principles of data sharing are becoming more widely accepted due to global recognition of the value of access and re-use. These principles need to be further embedded into research culture, through awareness raising, adoption of policies and incentive.

Researchers can minimally adhere to data access and sharing requirements by retaining their data, describing them appropriately and sharing them with others when requested. However, for widespread data sharing to occur, researchers need to be able to deposit their data into a repository where it will be maintained, curated

115 Private communication with Research Data Canada (October 10, 2014)

42

and can be discovered by others.

Data retention and preservation

In terms of full-scale preservation, these services are not yet widely available in Canada. There are some well-established government and domain repositories with preservation capacity. Scholars Portal in Ontario, the University of Alberta, and the National Research Council also maintain trusted digital data repositories.

As with access and sharing, researchers can provisionally adhere to data retention and preservation requirements by ensuring their data are stored and backed up appropriately.

Data management plans

Data management plans oblige researchers to describe how they will manage their data during the course of the research project and outline their plans for sharing and preserving data once the project is completed.

The main challenge of requiring data management plans is acceptance by the research community. Few researchers have an understanding of what a good data management plan entails. Researchers will need support for filling out data management plans.

The University of Alberta currently provides access to an automated tool, called DMP Builder, which assists researchers in developing data management plans. This tool is available to everyone and there will soon be a Portage version to offer support in both French and English. In the context of this tool, guidance is also being developed to help researchers understand and respond to requirements. More detailed expertise will also be available at some individual institutions and through the Portage network of expertise.

National capacity in this area will only develop over time, but raising awareness of the benefits of DMPs will be important to ensure their acceptance in the research community.

Canadian researchers can be expected to develop DMPs and their implementation will help build an understanding of data management planning in the research community.

43

9. Conclusion The global trend towards research data management and sharing is being driven by a number of things: to ensure verification of research results; to improve the quality and efficiency of research; and to promote the re-use of research for new discoveries and innovation.

This review found that, although there are many gaps and barriers, the environment for policy adoption for RDM in Canada is improving and Canada has made significant progress since the OECD declaration in 2004. There have been both bottom-up and top-down efforts to advance RDM infrastructure and expertise in Canada. Further targeted government investment and incentives could accelerate these advances.

It is clear that policies cannot be adopted in isolation. Good research data management practices depend on multiple factors being in place – including incentives, skills and expertise, services, infrastructure, funding, and policies. In addition, comprehensive adoption of good data management practices involves significant cultural change across numerous stakeholder groups. This will take time, and will likely progress through steady, incremental steps across multiple factors and domains. Parallel efforts must also be made to increase awareness and acceptance of policy objectives within the research community.

Despite the challenges, it is clear that policies are an extremely powerful lever to push the community forward. They provide a framework that helps to guide best practices and without them it is unlikely that there will be widespread adoption of RDM in Canada. Countries that have chosen to move ahead with policy implementation have found that although full compliance cannot be expected immediately, policies can greatly assist in raising awareness of RDM. As noted in a 2013 TC3+ consultation document, “Canada now stands in direct competition with a host of other countries, including the United States, European Union countries, Australia and other technologically advanced countries, in the race to develop an effective strategy for harnessing the digital wave.”116 RDM policies are an important component of any such strategy.

116 http://www.sshrc-crsh.gc.ca/about-au_sujet/publications/digital_scholarship_consultation_e.pdf