This document summarizes a presentation about meeting federal data sharing requirements. It discusses the history of these requirements and defines good practices for data sharing and stewardship. It also reviews some public data sharing services and provides tips for evaluating them. Key aspects of good data sharing include maximizing access, protecting privacy, ensuring proper attribution, and having long-term preservation and sustainability plans. The presenter emphasizes that restricted-use or sensitive data can be effectively shared through secure virtual environments.
1. From Data Sharing to Data
Stewardship: Meeting Federal
Data Sharing Requirements
ACRL 2015
Thursday, March 26, 2015
ICPSR – University of Michigan
Hashtag: #icpsr
5. Direct identifiers
• Addresses, including ZIP and other postal codes
• Telephone numbers, including area codes
Indirect identifiers
• Exact dates of events (birth, death, marriage)
• Detailed income
• Detailed geographic information (e.g., county)
6. “The study is composed of about 180,000 autopsy x-
ray image files taken of 58 corpses. The images
originally arrived on DVD and are formatted to
comply with the Digital Imaging and
Communications in Medicine (DICOM) standard….
The images are the data of the study, the images
files themselves contain metadata (metadata on the
images) scrubbed of identifiers but there isn't much
in terms of documentation.”
8. Today
• History (brief!) of federal data sharing requirements
• What is good data sharing? How do you achieve data
stewardship?
• Public data sharing services – tours & take-away tips
• Resources for creating data management plans and
funding quotes
9. You should leave this session with -
• Keen understanding of several sustainable data
sharing models
• Ability to assess data sharing services
– Through review of several services
– Walk-away tips for evaluating
• Knowledge (a portal) of resources for creating
data management plans for grant applications
10. • 50+ years of
experience
• Data stewardship
• Data management
• Data curation
• Data preservation
ICPSR
12. Recent Federal Data Sharing Initiatives
• NIH: 2003 – data sharing plans
• NSF: 2011 – data management plans
• OSTP: 2013 – Memo with subject “Increasing
Access to the Results of Federally Funded
Scientific Research”
19. Data Portion of Memo - 13 Elements
• The elements are also summarized online
within ICPSR’s Web site:
http://icpsr.umich.edu/content/datamanagement/ostp.html
21. UK results on data sharing attitudes
• In 2011 survey, 85% of researchers said they
thought their data would be of interest to
others.
• Only 41% said they would be happy to make
their data available.
• Only a third had previously published data.
Source: DaMaRO Project, University of Oxford
http://www.slideshare.net/DigCurv/15-meriel-patrick
22. Data Sharing Status
Federal
Agency
Shared
Formally,
Archived
(n=111)
Shared
Informally,
Not
Archived
(n=415)
Not
Shared
(n=409)
NSF
(27.3%)
22.4% 43.7% 33.9%
NIH
(72.7%)
7.4% 45.0% 47.6%
Total 11.5% 44.6% 43.9%
Pienta, Gutmann, & Lyle (2009). “Research Data in The Social Sciences: How Much is Being Shared?”
http://ori.hhs.gov/content/research-research-integrity-rri-conference-2009
See also: Pienta, Gutmann, Hoelter, Lyle, & Donakowski (2008). “The LEADS Database at ICPSR:
Identifying Important ‘At Risk’ Social Science Data.”
http://www.data-pass.org/sites/default/files/Pienta_et_al_2008.pdf
Pienta, Alter, & Lyle (2010). “The Enduring Value of Social Science Research: The
Use and Reuse of Primary Research Data”. http://hdl.handle.net/2027.42/78307
23. What is good data sharing - the basis of
data stewardship?
1.Maximize access
2.Protect confidentiality and privacy
3.Appropriate attribution
4.Long term preservation and sustainability
5.Data management planning
27. A well-prepared data collection
“contains information intended to
be complete and self-explanatory”
for future users.
Do no harm.
28. Protect confidentiality and privacy
• It is critically important to protect the identities of research
subjects
• Disclosure risk is a term that is often used for the possibility
that a data record from a study could be linked to a specific
person
• Data with these risks can be shared via a secured virtual
environment
• Data concerning very sensitive topics can also be shared via
a secured environment
29. Appropriate Attribution
• Properly citing data encourages the replication of
scientific results, improves research standards, guarantees
persistent reference, and gives proper credit to data
producers.
• Citing data is straightforward. Each citation must include
the basic elements that allow a unique dataset to be
identified over time: title, author, date, version, and
persistent identifier.
• Resources: ICPSR's Data Citations page , IASSIST's Quick
Guide to Data Citation, DataCite.
35. Data Management Planning
• Data management plans describe how researchers
will provide for long-term preservation of, and
access to, scientific data in digital formats.
• Data management plans provide opportunities for
researchers to manage and curate their data more
actively from project inception to completion.
• See ICPSR's resource: Guidelines for Effective Data
Management Plans
36. The Status of Data Sharing
– Good data sharing exists!
– Good data sharing requires funding -
sustainable funding!
– Sustainable funding for free public access
remains a challenge
37. Sustainable Data Sharing Models –
Three to Explore
• Fee for access model (subscription model)
• Agency model (agency or foundation funds
public access)
• Fee for deposit model (researcher writes fee
into grant and pays at deposit to fund public
access)
38. I. Fee-for-Access Data Sharing
• Funding is maintained by annual subscription fees charged to
institutions; individuals at subscribing institutions have free
(open) access to data
• Pooled (ongoing) subscriber fees are used to acquire, curate,
and maintain the service
• The service, open to everyone, is thus sustained by subscribers,
but agencies indicate these models are not ‘open enough’
because of the access fees
39. II. Agency-funded Data Sharing
• Agency sponsors/funds (ongoing) data curation & sharing enabling the
public to access without charge
• The archive is hosted with a curation entity like ICPSR where the public
can easily discover and access data and restricted-use data can also be
securely shared
• Agency directs data selection and compliance policies
40. III. Fee-for-Deposit Data Sharing
• Depositor (individual or entity) pays for data to be
curated and stored – a fee at deposit
• Deposit fees should be written into the grant
application
• Incoming deposit fees sustain the service and the
professionals behind it
• Sustainability risk fairly high in this model as it
depends upon:
– Continuous influx of deposit fees
– Depositors to put allocated fees towards curation & sharing
• Data tends to be bit-level (not curated): WIDIWYG
41. Fee for Deposit Services Arriving Daily!
(tips for evaluating coming shortly)
42. First: A Side-Note on Sharing
Restricted-Use Data
• Data with disclosure risk –
potential to identify a research
subject
• Data with highly sensitive
personal information
What is Restricted-Use Data?
43. Common Objection/Misperception:
“My data are too sensitive to share. . .”
• ICPSR has been sharing restricted-use data for
over a decade. Three methods are used:
– Secure Download
– Virtual Data Enclave
– Physical Enclave
• ICPSR stores & shares over 6,400 restricted-
use datasets associated with over 2,000
‘active’ restricted-use data agreements
44. Reality: Restricted-use data can be
effectively shared with the public
• Through the use of a virtual data enclave where
the data never leave the server
• Where there is a process (and understanding!)
to garner IRB approval from the requesting
scientist’s university
• Where there is a system, technology, data
professionals, and collaboration space in place
to disseminate (expensive to build!)
• Because agencies do allow for an incremental
charge to the data requestor to offset marginal
costs
45. Review of Public Data Sharing Services
• Overview of public data sharing services we have
reviewed
– Some key strengths of each
• Disclaimer: ICPSR has recently launched a public access
service (hosted)
– You’ll likely notice some bias when we talk about the
strengths of openICPSR
– And because we built the service, we know much more
about it
– Still, ICPSR’s public access service isn’t for everyone –
more on that shortly
48. How is openICPSR unique?
openICPSR is a public data-sharing service:
• Where the deposit is reviewed by professional data curators who
are experts in developing metadata (tags) for the social and
behavioral sciences = discoverable
• With an immediate distribution network of over 750 institutions
looking for research data, that has powerful search tools, and a
data catalog indexed by major search engines = usage
• Sustained by a respected organization with over 50 years of
experience in reliably protecting research data = sustainable
• Prepared to accept and disseminate sensitive and/or restricted-
use data in the public-access environment = protection of research
subjects
49. How will openICPSR disseminate sensitive
data to the public?
• The deposit of sensitive (restricted-use) data is similar
to the deposit of non-sensitive data except that the
depositor will indicate that the data should be for
restricted-use only
• Dissemination of sensitive data will be through
ICPSR’s virtual data enclave; in this environment, data
never leave the secure server and analysis takes place
in the virtual space
• Scientists desiring to access the data will need to
apply for the data and will pay an access fee
• openICPSR has already received sensitive (restricted-
use) and dissemination of these data has begun
50. openICPSR for Institutions and Journals
• Uses openICPSR platform
• Fully hosted in the ICPSR
cloud – no tech or patches
needed
• Branded with a logo and
colors
• Deposits incorporated into
ICPSR’s data catalog
• On-demand administrative
usage tools
51. A final note: openICPSR accepts research data from
a wide array of disciplines/fields, but not all
52. Tips for Evaluating a Data Sharing Service
• How will the service sustain itself? Does it have a long term funding
stream?
• How will the service care for my data in the long term should the service
fail? Is there a plan? A safety net?
• Can the service quickly maximize discoverability of my data? Does it
explain how it will do so?
• Does the service have a network of interested researchers & students
seeking data? Will my data get used?
• Does the service have knowledge of international archiving standards?
• Does the service provide a DOI, data citation, and version control should I
need to update my files?
• I have sensitive data or data with some disclosure risk to deposit. Does
the service understand how to secure it upon intake and when sharing?
Does it have experience in this area?
Questions to consider when selecting a data sharing service:
55. Purpose of Data Management Plans
• Data management plans describe how researchers
will provide for long-term preservation of, and
access to, scientific data in digital formats.
• Data management plans provide opportunities for
researchers to manage and curate their data more
actively from project inception to completion.
59. And still more guidelines after the
project is awarded:
• Guide emphasizes
preparation for data
sharing throughout
the project
• Available online and
via download (pdf)
60. ICPSR Data Curation Training Workshops
• 1-5 day workshops on data curation/data
repository management decisions
– Participants learn about best practices and
tools for data curation, from selecting and
preparing data for archiving to optimizing and
promoting data for reuse
• Available via ICPSR Summer Program (Ann
Arbor – July 27-31, 2015) or onsite at your
institution
61. Copies of these Slides & Use
• Feel free to share it; present
it; cite it!
• Find copies of these slides
on Slideshare.net
– Several notes and
additional links are found in
the notes view
62. Get More information
• Visit ICPSR’s Data Management &
Curation site:
http://www.icpsr.umich.edu/datamanage
ment/index.jsp
• Contact us:
– netmail@icpsr.umich.edu
– (734) 647-2200
• More on Assuring Access to
Scientific Data: white paper –
“Sustaining Domain Repositories
for Digital Data”
Notas del editor
Federal agencies are requiring data management plans as part of research proposals to increase public access to results (including research data) of federally funded scientific research. Join us for a session on sustainable data sharing models, including models for sharing restricted-use data. Demos of these models and tips for accessing hosted public data access services will be provided as well as resources for creating data management plans for grant applications.
Here’s the wave of ‘big data’.
Source of slide: Myron Gutmann’s IDF Meeting (June, 2007)
ICPSR exists to preserve and share research data to support researchers who:
Write research articles, books, and papers
Teach or utilize quantitative methods
Write grant/contract proposals (require data management plans)
Current archives/collections/repositories already meeting public access requirements regarding data
NACDA – NACJD – SAMHDA: examples of long term sustainability
NAHDAP – SAMHDA – DSDR: examples of sharing of confidential data
NACJD – example of depository/researcher compliance (holding 10% of funding to PI)
LGBT – MET: unique infrastructure and dissemination
Research Connections: reports and data dissemination; audiences including policymakers
In January 2011, the National Science Foundation released a new requirement for proposal submissions regarding the management of data generated using NSF support. All proposals must now include a data management plan (DMP). (NIH has similar DMP requirements.)
The plan is to be short, no more than two pages, and is submitted as a supplementary document. The plan needs to address two main topics:
What data are generated by your research?
What is your plan for managing the data?
The OSTP Memo
This memo directed funding agencies with an annual R&D budget over $100 million to develop a public access plan for disseminating the results of their research
concern for investment: “Policies that mobilize these publications and data for re-use through preservation and broader public access also maximize the impact and accountability of the Federal research investment.”
Federal agencies with over $100 M annually in R&D expenditures to develop plans to support increased public access to the results of research funded by the Federal Government
The OSTP Memo – Overview
Released February 22, 2013
A concern for investment: “Policies that mobilize these publications and data for re-use through preservation and broader public access also maximize the impact and accountability of the Federal research investment.”
Federal agencies with over $100 M annually in R&D expenditures to develop plans to support increased public access to the results of research funded by the Federal Government
“Maximize access, by the general public and without charge, to digitally formatted scientific data created with Federal funds…”
4,883 NIH & NSF PIs emailed a survey
1,217 responses (24.9% response rate)
1,003 valid (collected data, not disseratation)
We attempted to invite all 4,883 of these Pis.
The PI survey consisted of consisted of questions about research data collected, various methods for sharing research data, attitudes about data sharing and demographic information. PIs were also asked about publications tied to the research project including information about their own publications, research team publications, and publications outside the research team. We received 1,217 responses (24.9% response rate). For the analytic sample we select PIs and their research data if (1) they confirm they collected research data (86.6% of the responses), (2) they did not collect data for a dissertation award (n=33), or (3) they were missing data on the dependent variable.
Today we’ll talk about how to prepare your data collection so – and this is the ultimate litmus goal -- it “contains information intended to be complete and self-explanatory” for future users. [Quote is from the National Longitudinal Survey of Youth’s explanation of its documentation (see: http://www.nlsinfo.org/nlsy97/97guide/chap3.htm#threethree).]
Why does this matter? 1) Others will be able to independently use/understand data, 2) Data will be readable (i.e., in useable formats) in the future, 3) It makes your life less complicated once you’re finished with the data collection -- you don’t need to continually explain, reformat, revise, etc.
This isn’t rocket science, but it’s still important. I recognize that many watching this Webinar have extensive data backgrounds, so I’m going to convey the information as quickly and directly as possible.
Cuneiform tablet
EBCDIC format
Such is the dilemma! Good data curation is sustained by subscriber fees that pay for good documentation, data cleaning, rendering into accessible formats, preservation including file migration and production of ASCII files, and sustained storage and website delivery (and all of the data and tech professionals conducting that work). However, this model has been determined at this time not to be open ‘enough.’
This model is a fantastic solution to sustainability – providing the agency continues to allot funds for data curation and sharing. Access to the public is free with this model.
The agency pays organizations like ICPSR for data and tech professionals to process, preserve, and share their data. The agency sets the rules for what data it will include in its archive (sets the data selection policy) since not all data can/should be curated. The agency can also set rules for compliance to encourage researchers to deposit and share their data. One agency requires that ICPSR provide confirmation that data and documentation have been received and are in workable shape prior to releasing the final payment to the researcher. Unfortunately this is rare in terms of compliance, as it works!
The risk to sustainability is not zero however, since budget cuts to federal funding are always of concern.
The signals coming from federal agencies regarding public access as well as international momentum are resulting in a rush to provide solutions for open (free) public access to data.
Some entities are providing ‘open source’ solutions where an organization grabs the code & the tech group builds the repository out to its desire. (They’re expected to share this code back with the community later.) It makes sense that later these entities might share a catalog – otherwise, how these data will really get discovered (required for use of the data!), is a mystery.
Other entities are providing ‘fully hosted’ solutions where an organization or individual deposits data into the cloud (servers) of the hosting organization. This is the solution we’ll concentrate on when providing tips for evaluating data hosting services.
Sensitive personal information isn’t about names, addresses, credit card numbers, or other direct identifying information. Research scientists should never, never, ever submit this type of information to any hosted service – ever. What we’re talking about is highly personal information (topics) within research data that may include past/present drug use, illegal activities, or perhaps sexual habits.
We’re currently adding about 50 new agreements each month.
Figshare: $8-$15 per month per individual though public space free; funded by Digital Science out of London (funded by MacMillan Publishers); accepts all files types from all disciplines; provides DOI; will accept data from any discipline
Dryad: $80 per data package (individual) or bulk deposits from an institution for minimum $1,750 on up; also has member fees from $1,000 - $5,000, annually; focuses on data underlying international scientific and medical literature (replication); provides DOI
DataShare: There are at least two – one out of University of Edinburgh & one out of UC San Francisco – focusing on UCSF. Currently funded by the California Digital Library & closed to UCSF; however it plans to open up to other institutions in future phases; really nice interface and great presentation of Why you should share data! http://datashare.ucsf.edu/xtf/search?smode=stepsPage
DSpaceDirect: Product of Duraspace; $3750 to $8250 annually depending on the level of storage – targeted at institutional membership to sustain funding; set up so Google can discover content – search looks localized to the institution at this time (not yet a common catalog across DSpace entities); software provides DOI; will accept from any discipline
Dataverse Network: free deposits; funded by Harvard; accepts all files; provides DOI; is opensource and has 8 other sites using DVN – likely they share a catalog
Academic Torrents: distributed data repository where the focus is to accept really large (TBs) datasets; out of Umass Boston – has about 113 datsets as large as 84GBs; no fees found; looks like new entry attempting to offer a solution to big data!
There is significant administrative burden required for the dissemination of restricted-use data. This includes the completion and review of restricted-use contracts that include IRB approval, data protection issues, placement of the data into the VDE and monitoring of progress and results with a disclosure review of results as well as server time. This is what the access fee to the data user will cover.
openICPSR for Institutions and Journals was built to:
Fulfill an organization’s governmental grant & journal replication requirements
Brand the data-sharing service with your logo, colors, and a unique URL
Provide DOIs & data citations upon publishing
Increase exposure & reach of the organization’s research via inclusion in ICPSR’s data catalog & integration with your social media
Administer the fully-hosted (cloud) service economically without the need for costly technical staff or equipment
Share and preserve restricted-use data
Provide confidence that the data and service are safe & available for the long term
It is sometime easier to identify what openICPSR does not accept than what it does. openICPSR is not appropriate for the natural or hard sciences (bio-medical). It is also not appropriate for huge datasets – multiple GBs of data. Our meta-data experts and our catalog is focused on a very broadly defined area known as the social and behavioral sciences.
For repositories outside ICPSR’s domain, see Stanford’s list: http://library.stanford.edu/research/data-management-services/share-and-preserve-research-data/domain-specific-data-repositories
A collection of resources (links) to assist in data management plans for grant proposals
Tools to prepare plans (templates & sample plans)
Contact information for plan advice
https://dmp.cdlib.org/
Puts together the basic structure & form for your DMP. Note that it isn’t plug and go – the reasoning behind your management plans based on the discipline and/or data being collected should be added.
22 pages of guidelines and references even including a sample plan (boilerplate!) available for download.
Link to pdf document: http://www.icpsr.umich.edu/files/datamanagement/DataManagementPlans-All.pdf
Pdf link to the data prep guide: http://www.icpsr.umich.edu/files/deposit/dataprep.pdf
More information on data preparation for archiving: http://www.icpsr.umich.edu/icpsrweb/content/deposit/guide/