• All of the GWLA institutions who answered the survey (35 of 38), offer Research Data instruction services in the form of course-related instruction, workshops, and/or consultations.
• The top three topics for data workshops are general RDM (research data management) , specific tools, and programming languages.

• Over 85% stated that their institution had a repository that accepted research data, however 70% ofinstitutions have a single repository that is used for both data and other scholarly material. 30% have dedicated data repositories.
• 86.7% of repository policies address who is eligible to deposit, 73.3% address the type of files, 60% address the inclusion of sensitive date, 53.3% address the size of files, and 56.7% have a terms of deposit document.
• Institutions were split over mediated vs. unmediated deposit models, with 65.2% offering mediated approaches.
• Creative Commons licenses (CC-0 and CC-BY) are the most popular forms of licenses available in respondents repositories.
• 79% of respondents have repositories that produce persistent identifiers.
• Institutions were evenly split over using Software as a Service (SaaS) vs. self-hosted. DSpace and Digital Commons were the most popular software platforms used.
• 70% of libraries do not share staff across both their data and institutional repositories.
• Only 30% of repositories accept data that is still being updated.
• 80% of repositories allow for embargo periods. 75% of those have no limit on the length of embargo.

• 80% of GWLA libraries do not have policies that address research data. However, 57.1% have campus research data policies and 61.7% said that their library or institution has a strategic plan or mission that addresses research data.

This section only asked about groups unaffiliated with the library. There is likely overlap between services offered by libraries and other groups on campus. However, the survey did not ask for this information.

The most common RDS (research data services) offered by groups on campus unaffiliated with the library are:

• 65.7% - Statistical software support
• 65.7% - Data analysis support
• 62.9% - Active research data storage and backup solutions
• 60.0% - GIS and geospatial analysis
• 57.1% - Data visualization support
• 57.1% - Assistance locating data storage and backup solutions

The least common RDS offered by groups on campus unaffiliated with the library:

• 5.7% - Metadata assistance
• 20.0% - File organization and naming conventions
• 22.9% - Topic or How-To Guides
• 28.6% - Locating and using existing data; including identifying and suggesting repositories

The most common groups offering RDS on campus beyond the library are:

• Research and/or service centers/institutes (number of institutions=23)
• Information technology units (number of institutions=20)
• Academic departments or colleges (number of institutions=14)
• High Performance Computing units (number of institutions=13)

• The number of library staff dedicated to RDS ranges from 0 to 3.5 FTE (AVG 1.3 FTE). Meanwhile the number of staff who provide RDS, but are not dedicated to it, shows a much wider range (0-15). Together these two numbers indicate that how libraries staff RDS varies widely. The survey did not ask to differentiate between tenure track, permanent, or contract employees.
• Library staff who provided RDS fall within many different departments and job descriptions. Some institutions have dedicated RDS staff while many appear to expect library staff (such as liaison librarians) to preform RDS in addition to their other duties.
• Half of the libraries surveyed said they had a committee or group dedicated to RDS.

The survey was distributed in March and April of 2018. Invitations to participate were sent to the deans/library heads of the 38 GWLA institutions. Thirty-six of the thirty-eight institutions responded to the survey. However, two of the institutions only provided partial responses, one of which had to be discarded as only 12% of the survey was completed. The other partial response was complete enough to include in the majority of the analysis.

Analysis of the survey answers was only performed at the question level. This was done for two primary reasons. First, the survey was not designed for research or to preserve institutional or individual anonymity. The task was to find out what 38 institutions were doing and how they were doing it. Second, fact checking of major outliers (e.g. a reporting of 5000 library staff) indicated that self-reported statistics were found to be inaccurate and had to be discarded which prevented cross-analysis by library and parent institution demographics. The survey also did not yield sufficient data for any meaningful analysis for annual software and storage costs. Even with these restrictions, the authors believe that the data presented in this report are useful to other academic libraries who are exploring or building up research data services.

As the survey was not anonymous the authors have decided that access to raw, and most coded data, should be restricted to GWLA institution members. Only answers to questions which contain public information have been shared. Links to these data sets can be found within the text.

This section asked for contact information and library and parent institution employment numbers. Results from the demographics section show that GWLA members vary widely in size and staffing levels. However, the data gathered from survey respondents in this section contained multiple major outliers (e.g. 5000 FTE library employees or 38 FTE university faculty) which prevented analysis by institution or library size. An overview of the GWLA member institutions is available in Table 1. This data was not obtained through the survey and should only be used to gain a "ballpark" prespective of the membership.

This section asked respondents to provide information on their institution’s RDM teaching activities and services offered, as well as information on campus groups external to the library that offer RDM services.

Which Research Data Instruction services are offered at your institution? Table 2

Please supply titles or topics for the research data workshops your institution provides.

Answers from thirty-two institutions were analyzed (n=32). Workshop titles/topics were assigned up to 2 topic codes based on the information provided (Fig. 1). The most workshops/topics provided by an institution = 15 (one institution), the least workshops/topics provided by an institution = 1 (five institutions), and the average number of workshop/topics = 5. Workshops/topics devoted to specific tools or programming languages were coded and tracked separately from topic codes (Fig. 2). The coded data, codebook, and analysis for this question is available through Open Science Framework, see files labeled "Q08" in Murray et al. (2019)

Which of the following research data metadata services does your library provide? (see Table 3)

Please identify which, if any, research data services are currently provided by other groups on your campus. Table 4

If Research Data Services are currently provided by other groups on your campus institution please identify the groups offering the services.

Groups were assigned a type code based off the names and descriptions provided (Fig. 3). Groups that could be identified as belonging to a specific discipline were coded and tracked separately from the type codes (Fig. 4). The coded data, codebook, and analysis for this question is available through Open Science Framework, see files labeled "Q11" in Murray et al. (2019).

Library and institutional research data policies Table 5

Follow up questions asked for links to library and institutional policies if they were public. These are presented in Suppl. material 1.

Does your library's or institution's strategic plan or mission address research data services? (n=34)

• Yes, 61.8% (21)
• No, 38.2% (13)

This section asks respondents questions related to the scope and governance of their institution’s data repository.Thirty institutions responded to all the questions in this bloc, while 5 institutions (14.3%) indicated that they did not have a repository that accepted data and skipped to the last question bloc (Library Organization) .

Do you have a research data repository or a repository that accepts research data? (n=35)

• Yes, accepts research data = 30 institutions / 85.7%
• No repository or repository does not accept data = 5 institutions / 14.3%

Does your institution have a dedicated repository for research data or is the same platform used for both data and other scholarly materials? (n=30)

• Dedicated data repository = 30.0%
• No, same platform = 70.0%

Do your institutional repository and data repository system share staff? (n=30*)

• Yes, share staff = 30.0%
• No, do not share staff = 70.0%

* This question was accidentally given to all survey respondents, regardless of if they had a separate data repository or not.

Which of the following are addressed in the policies and/or information pages for the data repository? Table 6

Which of the following are addressed in your data repository’s policies and/or information pages? Table 7

Which "stages" of data does your data repository accept? Table 8

Are embargo periods available for deposited data? (n=30)

• Yes = 80.0%
• No = 20.0%

Does your institution limit how long data may be embargoed? Table 9

What is your library's preservation strategy for the data in the repository? Table 10

This section asked respondents to provide more technical details about their data repository’s operations, costs, and metadata capabilities. Only respondents that indicated that their institution has a repository that accepts data answered this bloc. For this section n=29 as one of the respondents did not complete this section of the survey.

What deposit model is used for research data? Table 11

For the purpose of this question, mediated was defined as "subject to review/changes and approval" and unmediated was defined as "no review or approval needed." The "other" choice was a free-text box. Two of the free-text answers indicated that there were plans to move to a mediated deposit model and one provided details on a hybrid model.

Which licenses are available? Choose all that apply. Table 12

Does your repository assign persistent identifiers to datasets? (n=29)

• Yes = 23 (79.3%)
• No = 6 (20.7%)

What type of software does your data repository use? (n=29)

• Software as a Service (SaaS), aka cloud-based software: 14 (48.3%)

• Self-hosted, aka local installation: 15 (51.7%)

What is the name of the platform (software) that the data repository runs on? Table 13

Two institutions reported using two platforms for their repositories. So, while the number of institutions responding to this question is 29, the number of repositories is 31.

Please provide the names of metadata schema(s) used to describe data the repository. Table 14

Table 14

Do library staff create or assist researchers with the creation of record metadata? (n=29)

For the purpose of this question: Record metadata is metadata that is searchable and harvest-able.

• Yes = 26 (89.7%)
• No = 3 (10.4%)

Do library staff create or assist researchers with the creation of documentation metadata? (n=29)

For the purpose of this question: Documentation metadata is metadata that exists to help others comprehend and reuse the data, such as a readme file.

• Yes = 18 (62.1%)
• No = 11 (37.9%)

This section addresses library staffing for Research Data Services and the job titles of library staff who provide RDS.

How many staff at your library provide research data services? Table 15

Answers from thirty-four institutions were analyzed (n=34). The most frequent number of staff providing RDS was 3 (reported eight times) while the average was 3.58. The highest number of staff was reported at 15 and the lowest was 0.5 for those institutions with staff that provide RDS.

Please provide an estimate of the combined Full Time Employee (FTE) dedicated to research data services in your library, accounting for the time of all staff involved. Table 16

Answers from thirty-four institutions were analyzed (n=34). The most frequent amount of combined FTE dedicated to RDS was 1.50 (reported by 7 times) while the average was 1.34. The highest amount of combined FTE was reported at 3.5 while the low was 0.10 for the institutions providing RDS.

Titles, positions, and departments of RDS staff.

This data was analyzed two different ways. Table 17 shows a word count analysis of the free text answers provided by the respondents while Table 18 analyzed the same data by assigning each position up to three codes.

Is there a Library Committee or Group dedicated to research data services? (n=34)

• Yes: 17 (50.0%)
• No: 17 (50.0%)