Download to read offline
Uploaded byLouise Corti
Use of data in safe havens: ethics and reproducibility issues
The document discusses ethics and reproducibility issues related to using data in safe havens. It summarizes the UK Data Service, which curates social science data and uses various safeguards to provide access to controlled data through its spectrum of access. It describes legal gateways for data sharing, the Digital Economy Act, and the UK Statistics Authority's accreditation process for researchers and projects. It also discusses the UK Statistics Authority's ethics self-assessment tool and factors that can impact reproducibility when data and code are behind access restrictions in safe havens.
More Related Content
Similar to Use of data in safe havens: ethics and reproducibility issues
Use of data in safe havens: ethics and reproducibility issues
- 1. Use of datain safe havens: ethics and reproducibility issues Louise Corti Service Director Data Publishing and Access UK Data Service UKRIO Research Integrity Webinar: Data Sharing and Ethics Online 11 November 2020
- 2. The UK DataService • Social science data service, funded by the UKRI ESRC • Curates and provides access to data for research and teaching • Trusted Digital Repository (TDR), accredited to ISO27001 Info Security Management standard, Digital Economy Act Processor application in process • Work closely with research funders and key data producers /institutions: research centres; UK NSIs, govt departments, British Library etc • Around 8000 data collections, from open to secure
- 3. Types of dataheld • One off large scale survey e.g. Health Survey for England Social survey • Detailed survey following people over time Longitudinal/cohort • Interviews, focus groups, diariesQualitative • UK census counts/tables, country level statistics Aggregate data • Digitised databasesHistorical data
- 4. Spectrum of Access:UK Data Service • No personal identification risk • Open licence; few restrictions on reuseOpen • Zero to low risk of personal identification • Authentication and authorisation Safeguarded • Risk of personal identification • Authentication and authorisation • Added safeguards Controlled
- 5. Safeguards for dataaccess Data access involves reduction of risk in a manner acceptable to the data owner Risk mitigated by legal gateways and appropriate safeguards 5 Safes Framework offers a portfolio of safeguards
- 6. Legal gateways toaccess • Government Departments, Local Authorities, agencies, public bodies face different legal restrictions on the nature of the access they can provide for research • In some cases, specific legal gateways have been drafted to facilitate data sharing • This can have an impact on the feasibility of some research projects and data linking between various data sources e.g. health and social data is often hard to link
- 7. • Digital EconomyAct (DEA) • Data Protection Act /General Data Protection Regulation (GDPR) • Statistics and Registration Service Act (SRSA) Gateways: Person-specific (‘fit and proper’) Project-specific Time-specific Dataset-specific • Disclosing the identity or identifying information to someone who should not have access is a breach of the law. Civil as well as criminal sanctions for breaches Legislation and legal gateways
- 8. Digital Economy Act,2017 • The DEA covers: Access to digital services, Digital infrastructure. Online pornography, Intellectual Property and Digital government • Has a useful Research strand that broadly enables de-identified information held by a public authority to be disclosed for the purposes of research in the public good • The research framework is underpinned by the Research Code of Practice and Accreditation Criteria, approved by the UK Parliament in July 2018 • The UK Statistics Authority is the statutory accrediting body • Research Accreditation Panel to oversee the independent accreditation of processors, researchers and research projects Digital Economy Act 2017
- 9. Digital Economy Actdata example
- 10. What makes thiscontrolled data? • Low level of geography • Date of birth, including day • Education and training: detailed training and qualifications • Household and family characteristics: detail on family and extended family units • Employment: industry code of main job, employment details • Unemployment and job hunting: details • Temporary leave from work • Accidents at work and work-related health problems • Nationality, national identity and country of birth • Occurrence of learning difficulty or disability • Benefits, including detail on type of benefits claimed
- 11. Controlled data Data access/availability • 5 SAFES framework (ONS) • Access only via an accredited (safe haven) • High access bar • Process for reproduction not set up by journals
- 12. 5 Safes frameworkfor Safe Data Access 12 SAFE PROJECTS + SAFE PEOPLE + SAFE SETTING + SAFE DATA + SAFE OUTPUTS = SAFE USE Is this an appropriate use of the data? How trustworthy are the researchers? Does the environment prevent misuse? Is the data detail appropriate? Is there any confidentiality risk from publication?
- 13. Balancing the FiveSafes: Example Adapted by the Australian Bureau of Statistics from ‘Five Safes: designing data access for research’, Desai, T, Ritchie, F and Welpton, R, 2016 Safe Unsafe 13
- 14. DEA Processors Accreditation process= Information Security + Capability https://uksa.statisticsauthority.gov.uk/digitaleconomyact-research- statistics/better-access-to-data-for-research-information-for- processors/
- 15. DEA Accredited Researchers Accreditationprocess = Safe Researcher Training + Test + Application https://uksa.statisticsauthority.gov.uk/digitaleconomyact-research-statistics/better-useofdata-for-research-information-for-researchers/list-of- accredited-researchers-and-research-projects-under-the-research-strand-of-the-digital-economy-act/ Understanding of disclosure risk + producing Safe Outputs Good attitudes and behaviour = Trust
- 18.
- 19. Ethics component • UniversityREC submission for projects undertaking secondary analysis of data often not always complete enough to pass ethics test; often focussed on data collection • Clear guidance designed by the National Statistician's Data Ethics Advisory Committee's (NSDEC) to support researchers and statisticians to complete an ethical self- assessment form
- 20. Ethics Self-Assessment • UKStatistics Authority Ethics Self-Assessment Tool • Easy-to-use framework to review the ethics of a project • Helps identify & mitigate any ethical issues • 6 main principles Public good, potential harm Identification, data security Training, technologies Legal gateways, frameworks Public view, engagement Access, use & sharing of data
- 21. Weightings for risk ‘Ethicalrisk’ defined as the ‘negative consequences of unethical actions’. Differential complexities of various ethical decisions assessed using weighted measures • Data linkage projects • Sensitive personal data and processing • Patient level health data • Research including children and vulnerable adults • Data sources
- 22. The Ethics Self-Assessmentform https://uksa.statisticsauthority.gov.uk/about-the-authority/committees/national-statisticians-data- ethics-advisory-committee/ethics-self-assessment-tool/ • For each item – provide a score • Enter the score in the spreadsheet • The spreadsheet calculates an overall score of risk level
- 23. Reproducibility factors insafe havens Impact of the research outputs is a key requirement Code can be taken out, but is reviewed for risk ✘ Hard to reproduce as data and code behind a big gate ✘ Reproducer needs Accredited Researcher status ✘ Requires mechanism for reproducer to be added to an accredited project ✘ Difficult provenance chain for ‘research-ready’ admin data (creation, cleaning and versioning) ✘ Journals not ready to enable this review process
- 24. Using the 5Safes network Strong 5 Safes framework for access to unsafe data Very simple ethics assessment – easy to mark Utilise Accredited Researchers to undertake reproducibility work Better training on how to be reproducible Advocate cleaning, value-added, final code available in safe haven to enable reproduction and reuse
- 25.