Information is Power: Information is Power: Overcoming Obstacles Overcoming Obstacles

to Data Sharingto Data Sharing

Professor Denise Professor Denise LievesleyLievesleyHead of School of Social Science and Public Head of School of Social Science and Public

Policy, Policy, KingKing’’s College Londons College London

and President, and President, International Statistical InstituteInternational Statistical Institute

Sharing data Sharing data –– two important publicationstwo important publications

Fienberg S., Martin and Fienberg S., Martin and StrafStraf (1985) (1985) ‘‘Sharing Sharing research dataresearch data’’ National Academy Press National Academy Press ArzbergerArzberger P., Schroeder, Beaulieu, P., Schroeder, Beaulieu, BowkerBowker, , Casey, Casey, LaaksonenLaaksonen, Moorman, , Moorman, UhlirUhlir, , WoutersWouters(2004) (2004) ‘‘Promoting Access to Public Research Promoting Access to Public Research Data for Scientific, Economic, and Social Data for Scientific, Economic, and Social DevelopmentDevelopment’’ Data Science JournalData Science Journal

Statistical datasetsStatistical datasetsData produced for other purposes (often Data produced for other purposes (often administrative or management)administrative or management)Research dataResearch data

Sharing Statistical dataSharing Statistical data

Aim Aim –– to encourage the widest to encourage the widest possible possible informedinformed use of data use of data consistent with the responsibilities consistent with the responsibilities with respect to confidentiality etc with respect to confidentiality etc

Collect once, use many timesCollect once, use many times

Benefits to data providers Benefits to data providers of sharing data with the of sharing data with the

research communityresearch communityDevelopment of knowledgeDevelopment of knowledgeEncourage greater exploitation of dataEncourage greater exploitation of dataContribute to sound policy decisionsContribute to sound policy decisionsFoster multiple perspectives on dataFoster multiple perspectives on dataFacilitate comparative researchFacilitate comparative researchCreate knowledgeable data communityCreate knowledgeable data communityProvide feedback on data and improve data Provide feedback on data and improve data qualityqualityImproving teaching and ensuring relevance to Improving teaching and ensuring relevance to official statisticsofficial statistics

Reduction of response Reduction of response burdenburden

Compliance costs important especially in Compliance costs important especially in small countries and in surveys of elites, small countries and in surveys of elites, businesses, institutionsbusinesses, institutionsFresh data collection takes time and Fresh data collection takes time and resourcesresourcesSecondary data analysis can take place in Secondary data analysis can take place in resource resource ––constrained environment constrained environment

There is growing awareness that failure to There is growing awareness that failure to exploit the full potential of official data has exploit the full potential of official data has costs for society and many official agencies costs for society and many official agencies now espouse the aim of ensuring that data now espouse the aim of ensuring that data are used as extensively as possible. are used as extensively as possible. For purposes of public accountability it is For purposes of public accountability it is important that official data are made important that official data are made available.available.Often these are data which the research Often these are data which the research community could NOT collect themselvescommunity could NOT collect themselves

Sharing administrative dataSharing administrative data•• Unrivalled & untapped level of detailUnrivalled & untapped level of detail•• Survey data has limitationsSurvey data has limitations•• Administrative data may have full Administrative data may have full coveragecoverage•• and better temporalityand better temporality•• Reduces respondent burdenReduces respondent burden•• Has potential cost benefitsHas potential cost benefits•• Opportunities for data linkage with other Opportunities for data linkage with other sourcessources•• Local ownership and involvementLocal ownership and involvement

Sharing research dataSharing research data

““Publicly funded research data are a Publicly funded research data are a public good, produced in the public public good, produced in the public interest. As such they should remain in interest. As such they should remain in the public realm. Availability should be the public realm. Availability should be restricted only by legitimate restricted only by legitimate considerations of national security considerations of national security restrictions; protection of confidentiality restrictions; protection of confidentiality and privacy; intellectual property rights; and privacy; intellectual property rights; and timeand time--limited exclusive use by limited exclusive use by principal investigators.principal investigators.””

Scientific paradigmScientific paradigm

The ISI declaration on professional ethics The ISI declaration on professional ethics states that states that ““A principle of all scientific work is A principle of all scientific work is that it should be open to scrutiny, assessment that it should be open to scrutiny, assessment and possible validation by fellow scientists.and possible validation by fellow scientists.””One of the fundamental principles of scientific One of the fundamental principles of scientific scholarship is that research findings together scholarship is that research findings together with the underlying data should be available for with the underlying data should be available for others to confirm, refute, clarify or extend the others to confirm, refute, clarify or extend the findings. findings. Promote deliberate replication, avoid ignorant Promote deliberate replication, avoid ignorant duplicationduplication

““In recent years, the debate on eIn recent years, the debate on e--science has science has tended to focus on the tended to focus on the ““open accessopen access”” to the to the digital digital output output of scientific research, namely, the of scientific research, namely, the results of research published by researchers as results of research published by researchers as the articles in the scientific journals. This focus the articles in the scientific journals. This focus on publications often overshadows the issues of on publications often overshadows the issues of access to the access to the input input of research of research -- the research the research data, the raw material at the heart of the data, the raw material at the heart of the scientific process and the object of significant scientific process and the object of significant annual public investments. In terms of access, annual public investments. In terms of access, availability of research data generally poses availability of research data generally poses more serious problems than access to more serious problems than access to publications.publications.”” ArzbergerArzberger et al (2004)et al (2004)

Incentives in academic systemIncentives in academic system

In 1985 the report of the US committee of In 1985 the report of the US committee of national statistics pointed out that national statistics pointed out that ‘‘A scientist is A scientist is recognised and rewarded through the scientific recognised and rewarded through the scientific community and its institutions. Researchers will community and its institutions. Researchers will have greater incentives to share data if the have greater incentives to share data if the community and its institutions foster the idea community and its institutions foster the idea that the practice advances science and is part of that the practice advances science and is part of what is recognised as necessary and proper what is recognised as necessary and proper scientific behaviourscientific behaviour””..Competition, performance targets, etcCompetition, performance targets, etcRole of the Research Assessment ExerciseRole of the Research Assessment Exercise

Barriers to data sharingBarriers to data sharing

Confidentiality and sensitivity of dataConfidentiality and sensitivity of dataLegal restrictionsLegal restrictionsPromises made to respondents Promises made to respondents Concerns about misuse of dataConcerns about misuse of dataEnsuring equity of accessEnsuring equity of accessNeed for revenue generationNeed for revenue generationAmbiguities over data ownershipAmbiguities over data ownershipConcerns about data qualityConcerns about data quality

Responsibilities of data usersResponsibilities of data users

acknowledge and give creditacknowledge and give creditrespect conditions of accessrespect conditions of accessprovide feedback on useprovide feedback on useensure the quality of their analysisensure the quality of their analysisavoid bringing the data providers into avoid bringing the data providers into disreputedisrepute

Value and role of data intermediariesValue and role of data intermediaries

Importance of establishing Importance of establishing policies on data access, policies on data access, sharing and preservationsharing and preservation

of official agenciesof official agenciesfunding bodiesfunding bodiesuniversitiesuniversitiesprofessional societiesprofessional societies

Example policy Example policy (UK Economic and Social Research (UK Economic and Social Research

Council)Council)

restricts new data collection,restricts new data collection,encourages secondary analysis, encourages secondary analysis, requires deposit of new data and derived requires deposit of new data and derived data in UK data archive, data in UK data archive, sets standards for documentation, sets standards for documentation, provides resources for data access and provides resources for data access and preservation, preservation, supports training of users,supports training of users,builds data commons.builds data commons.

Access Access –– one size doesnone size doesn’’t t fit allfit all

Needs of users/usages differNeeds of users/usages differ•• especially in relation to their sophistication and the need for especially in relation to their sophistication and the need for

individual level dataindividual level dataDiverse data sets especially in relation to Diverse data sets especially in relation to sensitivity of content and possibility of disclosuresensitivity of content and possibility of disclosureIntegrated, longitudinal and spatially Integrated, longitudinal and spatially disaggregated data pose particular challengesdisaggregated data pose particular challengesSo do administrative data So do administrative data •• Good practices exist for survey data but not for admin. data Good practices exist for survey data but not for admin. data

And crossAnd cross--national datanational data•• European social surveyEuropean social survey

PreservationPreservation

Having collected data at some cost to society, it Having collected data at some cost to society, it behoves us to manage them well. behoves us to manage them well.

Alongside dissemination, this entails data preservation.Alongside dissemination, this entails data preservation.Due to poor data management, human error as well as Due to poor data management, human error as well as

technical change and inadequate use of technical change and inadequate use of technology, many data sets are no longer technology, many data sets are no longer readable. readable.

Thus all that remains of this important legacy are the, Thus all that remains of this important legacy are the, often quite superficial, reports that were produced often quite superficial, reports that were produced at the time. at the time.

To this extent an important part of our heritage is lost To this extent an important part of our heritage is lost and we will be severely limited in our analysis of and we will be severely limited in our analysis of change. change.

MetadataMetadataIt is necessary not only to preserve data but also to It is necessary not only to preserve data but also to create and preserve metadata and contextual create and preserve metadata and contextual information. This is essential to ensure that the information. This is essential to ensure that the interpretation of the data will be informed. interpretation of the data will be informed. The documentation should includeThe documentation should include

data collection instruments and formsdata collection instruments and formsinstruction manualsinstruction manualsdefinitions and conceptsdefinitions and conceptsdescriptions of scope and coverage and other aspects of descriptions of scope and coverage and other aspects of qualityqualitycodebookscodebooksbasic tablesbasic tablesrecords of validation checksrecords of validation checks

Case studyCase study–– building the building the secondary uses services secondary uses services

National Health Service in EnglandNational Health Service in Englandindividual patient care recordsindividual patient care records

-- The collection of data which records The collection of data which records every interaction with the health every interaction with the health service from conception to autopsyservice from conception to autopsy

Two committeesTwo committees

One on potential usageOne on potential usage

–– Conducting audits of clinical practice;Conducting audits of clinical practice;–– Surveillance of infectious diseases Surveillance of infectious diseases –– Management of the health systemManagement of the health system–– Monitor equity of access and provision; Monitor equity of access and provision; –– EvidenceEvidence--based health policy based health policy –– Providing better information to the general Providing better information to the general

public public –– Improving the quality and safety of careImproving the quality and safety of care

Second committee on Second committee on governancegovernance

Hierarchy of data access consistent with Hierarchy of data access consistent with ensuring lowest risk of patient identification ensuring lowest risk of patient identification Need to knowNeed to knowRole of honest brokers and safe havensRole of honest brokers and safe havensDevelopment of Development of ‘‘virtualvirtual’’ safe havenssafe havens

Information governance of Secondary Information governance of Secondary Uses ServiceUses Service

aggregate data widely availableaggregate data widely availabledefault default anonymisedanonymised-- or or pseudonymisedpseudonymisedif identifiers needed consent should be if identifiers needed consent should be obtainedobtainedfull justification in terms of benefits to be made full justification in terms of benefits to be made for exceptionsfor exceptionsexceptions assessed by transparent, equitable, exceptions assessed by transparent, equitable, replicable and open process involving patients replicable and open process involving patients representativesrepresentativesrequirement for safety and security of requirement for safety and security of information (information (ieie accountability)accountability)

Concluding remarksConcluding remarksWe create a diverse range of datasets, many We create a diverse range of datasets, many of which are unique, rich in information content of which are unique, rich in information content and incapable of replication. and incapable of replication. Sharing allows scientists to extend the value Sharing allows scientists to extend the value of these datasets through new, high quality, of these datasets through new, high quality, ethical research and exploitation. It also ethical research and exploitation. It also reduces unnecessary duplication of data reduces unnecessary duplication of data collection. collection. Building preservation and documentation Building preservation and documentation systematically into routine data management is systematically into routine data management is part of good practice: it strengthens quality, part of good practice: it strengthens quality, enables replication and audit, and provides a enables replication and audit, and provides a sound basis for data sharing.sound basis for data sharing.