The Realities of Academic Data Sharing (RADS) Initiative, funded by the US National Science Foundation (NSF #2135874), is an exploratory study examining the costs, activities, services, and technical infrastructure to support data sharing of federally funded research at six academic institutions. In order to determine what activities are actually supported and the costs incurred, RADS has undertaken two research streams as part of the initiative: one involving funded researchers and one involving campus administrators. This update concerns the second RADS research stream, and highlights the data-sharing activities reported by campus administrators at the six participating RADS institutions: Cornell University, Duke University, University of Michigan, University of Minnesota, Virginia Tech, and Washington University in St. Louis.
To identify possible participants for the Institutional Infrastructure Survey, RADS principal investigators (PIs) each did a scan of their institution to identify the departments, units, or offices that support researchers in their data-sharing efforts. After this scan, administrators of these departments/offices were identified, and then invited to participate in the survey. In addition to this initial identification, other criteria for survey participation included: knowledge of department/office infrastructure expenditures; knowledge of personnel activities to support data sharing; and knowledge of personnel salaries. The number of identified offices/administrators varied among the six RADS institutions, from 15 to 34 offices; the response rate of the administrators also varied, from 29.4% to 76.5% (see Table 1 below). In the survey, administrators were then asked specifically if their department or office supported at least one of the 27 public-access data-management-and-sharing activities identified by the RADS team. These data-management-and-sharing activities occur across five research or grant life cycle phases, and are described in detail in the RADS Research Update #2 blog post.[1]
Table 1: Response rate of administrators invited to complete the RADS Institutional Infrastructure Survey
| RADS Institution | Invited Administrators | Number of Responses | Response Rate |
| Cornell University | 17 | 13 | 76.5% |
| Duke University | 15 | 9 | 60.0% |
| University of Michigan | 23 | 15 | 65.2% |
| University of Minnesota | 33 | 18 | 54.5% |
| Virginia Tech | 17 | 8 | 47.1% |
| Washington University in St. Louis | 34 | 10 | 29.4% |
| Total | 139 | 73 | 52.5% |
The 27 data-management-and-sharing activities outlined in the Institutional Infrastructure Survey serve as the foundation for a mapping, or visualization, of data-management-and-sharing activities by 1) service areas across all of the six institutions and 2) within each of these institutions separately. As such, there are two corresponding sets of visualizations (one set for service areas and one set for each institution). Furthermore, while we aim to present a robust mapping of data-management-and-sharing activities across each service area and across each institution, we recognize that, due to a lack of input from all identified departments/offices, our mappings may be incomplete. At the institutional level, our mapping efforts may be considered a starting point for further exploration and, at the service level, our mappings may be considered a broad-view depiction of which data-management-and-sharing activities are supported in each general area across our institutions.
For the service-level mappings, responding offices were grouped into four main categories:
- Campus libraries
- Information technology (IT) offices, including sub-units
- Administrative research offices, such as the Office of Research, the Office of the Vice Chancellor for Research, and Sponsored Projects Offices
- Academic institutes, research centers, or departments
We chose to group responding offices into these four broad categories to enable a mechanism for comparison across the six RADS institutions, all of which house different departments and/or offices to support researchers in data-management-and-sharing activities, and provide varying levels of service. Instead of attempting comparisons across institutions at an apples-to-oranges level, the aim of classifying the offices into these four broad categories is to enable apples-to-apples comparisons, as academic institutions typically have similar higher-level administrative infrastructures, even if the scale differs across institutions. Note: This Service-Area Legend lists all participating offices at each of the institutions and their service-area classification.
The visualizations below (Visualizations 1–4) represent the four service areas, across all six RADS institutions, and show where services are provided for each of the 27 data-management-and-sharing activities asked in our survey to administrators. To view which office at which institution supports a particular activity, hover over each rectangle to view the full activity name and the supporting office. Note: This Tableau Label Legend lists all Tableau labels with their corresponding activity.
Visualization 1: Data-management-and-sharing activities supported by campus libraries across the six RADS institutions
From the Libraries visualization (Visualization 1 above), we can clearly see that libraries support data-sharing activities, in some capacity, throughout the entire research life cycle. As key partners at academic institutions to support the research enterprise, libraries may offer services supporting upwards of two dozen activities to enable data sharing, including: consulting on data-use agreements, supporting intellectual-property questions, fielding copyright considerations, managing active data, providing data-curation-and-preservation services, and making data broadly available through infrastructure, such as institutional repositories (or providing consultation on alternative repositories). In contrast, libraries showed the least amount of support in the Closeout and Compliance phase of the research life cycle; however, one in six of our institutions (the University of Minnesota) did indicate support in this area as well.
Visualization 2: Data-management-and-sharing activities supported by Information Technology (IT) offices across the six RADS institutions
 |
Congruence with recent COGR research |
| COGR’s May 2023 report “The Cost of Complying with the New NIH DMS Policy” presents survey results and corresponding analysis showing that in all areas of the data life cycle, campus libraries and IT offices have the greatest burden at the administrative level, thus, garnering the highest costs for compliance at the administrative level. This corresponds to RADS findings that campus libraries and IT offices offer the most support, in terms of service, infrastructure, and personnel support, across the research life cycle. | |
Visualization 3: Data-management-and-sharing activities supported by administrative research offices (for example, Office of Research, Sponsored Projects Office) across the six RADS institutions
Visualization 4: Data-management-and-sharing activities supported by Research Institutes, Academic Departments, Institutional Centers, and related centers, across the six RADS institutions
Of particular interest to RADS researchers is how many data-sharing services are coming out of the specialized areas on their campuses (Institutes and Centers—see Visualization 4), as compared to what might be considered central services. This distribution of services may give libraries insight when considering how to support these data-sharing efforts at their respective institutions, and how important these organizations are to their disciplinary communities. Further research would be needed to determine what is driving the growth in these specialized institutes and centers, and if support was missing from central services, so that the centers and institutes filled that gap, or if it was the other way around.
Institution Spotlight: Cornell University Campus Mapping
Cornell University Library has a long and established history of providing research-data services and support. eCommons, the Cornell University institutional repository, launched in 2002, was among the first institutional repositories in the United States that also published research data.
To ease faculty burden and facilitate research-data services across distributed offices, Cornell University Library, has taken a leadership role in coordinating institutional services for research data with the launch of their cross-institutional research-data collaborative, the Research Data Management Service Group (RDMSG). Launched in 2010, this group is composed of consultants located throughout the university who are available to meet with faculty and researchers on their research-data needs. This group is coordinated through the Cornell University Library (CUL), with oversight by administrators from central IT, research computing, Weill Cornell Medicine, and faculty.
While Cornell University Library has a strong understanding of services for research data across campus, the RADS project provided an opportunity to dig deeper across campus to identify and understand the extent of services. Utilizing the RDMSG as a starting point, co-PI Wendy Kozlowski, library lead for research data services, first completed a website review of campus offices that she felt may be providing research-data services. Leveraging this list, she reached out to her RDMSG colleagues to review and identify any known office gaps.
After completing the survey and round of interviews, CUL data staff were surprised by the number of offices that did not indicate they provided data-sharing activities, such as the University Council and department-based IT services. Another big takeaway from the data-collection effort and analysis was the need for increased awareness of what comprises data sharing. Numerous responses indicated that while research-data security was a service they provided, they did not consider that service to be a component of research-data sharing.
Reflecting on the library services, CUL has invested a significant amount of resources and infrastructure around curation, especially since the original 2013 Holdren memo was published. Kozlowski notes, “We’ve observed a shift in needs for research data management planning. We’re receiving fewer requests on data management plans, and more requests for consultations on how to share research data.” Finally, she shares a final quick tip for librarians or those in a research consultancy capacity:
Ask for a copy of the data management plan (DMP) if researchers are requesting a consultation on how to share research data—it should help with the consultation and also facilitate a conversation on DMP updates and compliance.
Visualization 5: Data-management-and-sharing activities supported at Cornell University across all four service areas
Acknowledgements
The RADS team would like to thank Mollie Webb, data and GIS developer, University Libraries, Washington University in St. Louis, for the creation of the Tableau visualizations. We would also like to thank Lizhao Ge, RADS statistical consultant, for her work on the development of the Institutional Infrastructure Survey and corresponding analysis. The lightbulb icon used in this post was made by Good Ware from www.flaticon.com.
[1] Activities to describe data sharing across campus are also detailed in the November 2022 RADS report, Public Access Data Management and Sharing Activities for Academic Administration and Researchers.
