EUROSTAT
Directorate B: Quality, methodology and information systems
Unit B-2: Methodology and research /
ESTAT/B2/EGSDC(11)1
Doc.1 – V.2011.04.08
Available in EN only
1st meeting of the
Expert Group on SDC
Luxembourg, 5 and 6 April 2011
BECH building, Room B2/464
Report and conclusions from the meeting
1
1st meeting of the
Expert Group on Statistical Disclosure Control
Luxembourg, 5 and 6 April 2011
Report and conclusions from the meeting
Item 1 Welcome and adoption of the Agenda
Rainer Muthmann, Head of Eurostat Unit B2 Methodology and Research welcomed the participants and opened the meeting.
Item 2 Mandate of the Expert Group
The mandate, as presented at the last meeting of the Working Group on Statistical Confidentiality (WGSC) was discussed.
The following issues were stressed:
Strong commitment from the experts required
Emphasis on deliverables: discussions and meetings should lead to tangible results
Clear definition of actions and objectives
Avoidance of conflict of interests (the group advices Eurostat on specific actions but formal evaluation is under the responsibility of Eurostat; the group contributes to the assessment of results and make proposals to the WGSC for endorsement)
The need to allow some flexibility to the work programme and to take into account emerging needs
The need for good communication channels with other groups (especiallyWGSC) and projects
Few comments were made to the mandate. The revised version of the mandate is enclosed in the annexes.
Item 3 Work programme for 2011-2012
The objectives to be achieved in the area of microdata and tabular data protection were thoroughly discussed and milestones were refined. The revised version of the work programme is enclosed in the annexes.
The method of work was discussed. The principal function of the group is to make recommendations for the work on SDC in the ESS and to assess the work having been done. The group will also propose the WGSC to formally endorse documentswith ESS methodological relevance. The group has no resources to carry out the actual work on their own. It is the responsibility of Eurostat to find the appropriate means to implement recommendations of the group and to develop corresponding actions (ESSnet, service contract etc.)
The revised versions of the mandate and work programme will be submitted to the WGSC for final approval.
Item 4 Confidentiality impact level – formulation of the criteria
Item 5 Anonymisation of Community Statistics on Information Society (CSIS)
The document on Operationalisation of Confidentiality Impact Levels (room document) was discussed together with the document on anonymisation of CSIS, the latter serving as a case study of the approach proposed.
The Group expressed concerns about the possibility to develop a systematic approach to CILs. Further cases studies with more complex surveys will have to be developed. Eurostat will further reflect on the relevance of CILs and respective proposals for the future work in this field will be submitted to the Expert Group at a later stage.
On the specific issue of the protection of CSIS micro data, the group mentioned that the methodology of sampling design and sampling weights should have been taken into account in the analysis. It was however recognised that actual impact on the CSIS is likely to be limited given the already high coarsening of key variables.
Item 6 Guidelines on anonymisation of social surveys
The guidelines aim to implement the methodology developed in the handbook on SDC and to apply it on a particular case of EU social surveys Guidelines tends to promote common approach to the coarsening of variables taking benefit of the existence of harmonised list of (core) variables.
While recognising that the core variables are inmostcases the re-identification variables, the group insisted on not limiting the approach to these core variables and to talk about identification variables instead.
The groups stressed the need to make reference whenever possible to existing material in the handbook.
Some delegates suggested that the guidelines shouldjust recommend some standards such as e.g. standards for recoding of core variables. Moreover, emphasis should be given to explanationof consequences of different actions proposed.
The group is looking forward to the development of the new framework for harmonising micro data protection as currently tested in the ESSnet on SDC harmonisation.
Furthermore, it was underlined that the guidelines for anonymisation should address not only the SDC issues but also the governance frameworks of the process. Participants shared experiences of such governance frameworks in their countries. The following components of the process were mentioned:
Discussion with potential users
Discussion with sectoral units dealing with a survey in question
Discussion with legal unit
Final decision taken by a panel of experts (lawyers, methodologists, others)
Action plan:
Youri Baeyens and Anco Hundepool volunteered to work further on the guidelines. The document will be revised and a proposal for the governance structure will be made.
Item 7 First discussion on the development of public use files (CSIS case)
The experts underlined the relevance of the workon PUFs at the ESS level. In some countries this work is already on going and in some others the requests for PUFs are likely to grow.
The need for a clear definition of the potential users of PUF was stressed. Two different objectives can be distinguished:
1)To train potential users on the use of micro data files. Prevalence should be then given to the structure of the file with minimum requirements on the information content.
2)To foster scientific use of PUF putting more emphasis on information content (as it is a case in the US)
The different approaches will have to be tested. Some possible methods(reduction of categories, perturbation methods, synthetic files, subsampling) were briefly discussed during the meeting.
Action plan:
A document will be drafted for the discussion at the level of the WGSC. It should define the use of PUF and proposethe frameworks for further work at the ESS level. It may also contain review of the existing methods and practices in different countries. Keith Spicer and Christopher Guerke will be leading the work on this file. It was proposed to associate the scientific community to this task. A possible person could be Christopher Wolff (GESIS).
8. Ongoing SDC projects / events
The following projects have been presented:
- ESSnet on harmonisation of SDC
- Anonymisation of Farm Structure Survey
- Data Without Boundaries (DWB): FP7 project
- Expert Group for international collaboration on microdata access
- UNECE/Eurostat work session on statistical data confidentiality
The links to the projects' websites will be uploaded on WIKI.
It was also agreed that there will be a repository of documents relevant to the work of the group. To start with, it was decided to collect information on actual cases of attacks on databases.
Conclusions
The meeting was successful to clarify the mandate of the group and some organisational issues, to refine the work programme and define first actions. In the short-term horizon the guidelines on anonymisation of social surveys will be reviewed and the framework document on PUFs will be drafted for further consultations at the level of the Working Group on Statistical Confidentiality.
The issue of Confidentiality Impact Level concept remains open. The approach will have to be re-discussed by the TF on revision of regulation 831/2002 and more pragmatic solution will need to be considered.
The first meeting was focused on the micro data. The next meeting of the Group is scheduled at the end of November 2011 and will be devoted to tabular data protection.
Annexe 1
Participants list
BE / Mr. Youri Baeyens /CZ / Mr. Ivan Stojka /
DE / Mr. Christopher Guerke /
FR / Mr. Thomas Denoyelle
Mr Julien Nicolas /
IT / Ms. Luisa Franconi
Ms. Daniela Ichim / Excused
HU / Mr. László Antal /
NL / Mr. Anco Hundepool /
AT / Mr. Bernhard Meindl /
PT / Mr. Pedro Campos /
FI / Ms. Janika Tarkoma /
UK / Mr. Keith Spicer /
Estat / Jean-Marc Museux
Aleksandra Bujnowska
Mihaela Agafitei /
Annex 2
Mandate of the Expert Group on SDC
Name of the Group: Expert Group on Statistical Disclosure Control
Origin: Eurostat proposal
Objectives:
The objective of the Expert Group is to issue guidelines, recommendations and advice for the WGSC and propose methodological solutions, scientifically sounded for statistical confidentiality in the ESS. The group should strive for harmonisation of SDC in the ESS. The group will report to the WGSC. The WGSC will consult theExpert Group ad hoc on various issues related to protection of microdata and tabular data. The work programme (enclosed) agreed by the group will be submitted to the WGSC for validation. The informal and free flow of ideas will be favoured. Scientific consensus should be sought in the recommendations issued. Conflict of interest should be avoided.
Tasks:
- coordinate the development of methodology and tools for SDC in the ESS
- peer review of the results of the methodological projects related to SDC (ESSnet, service contracts related to statistical confidentiality)
- analyse the methodology for the risk of disclosure of microdata released under Regulation 831/2002
- make recommendation for the protection of EU microdata to be released under the different modes (public files, anonymised micro data, remote access, safe centre access …); validate the disclosure control methods developed
- validate guidelines, handbooks and methodological frameworks aiming at harmonising SDC in the ESS
- monitor/control the development of SDC methods and tools;
- identify opportunities for cooperative methodological development in SDC in the ESS in view of their possible harmonisation
- issue recommendation for the training and competence development in SDC in the ESS
- maintain a repository of standards, recommendations and good practices and methodological documents for SDC
Frequency: Meetings are being convened regularly, normally twice a year, frequent virtual communication favoured.
Representation: SDC experts proposed by the WGSC and nominated by Eurostat.
The representatives should have practical expertise and experience in SDC methods and tools. The optimal number of representatives would be around 10. The Expert group is not a forum for exchange of good practices.
Tools of communication: E-mail contact, CIRCA, WIKI.
Related structures: Working Group on Statistical Confidentiality, Directors of Methodology Working Group, different thematic Working Groups and Sector Groups
Chair: Eurostat – Unit B2 Methodology and Research
Secretariat: Eurostat – Unit B2 Methodology and Research
Annexe 3
Work programme for 2011 and 2012
Microdata access
- High priority:
- Formulation of the criteria for CI levels: deadline: Sept. 2011
- Medium priorities:
- Initiation of work on public use files: end 2012
- Developing online access to Eurostat micro databases: end 2012
- Proof of concept of a new model for harmonisation of micro data protection (ESSnet project): beginning 2012
- Low priority:
- Revision of guidelines for social microdata anonymisation: 2011 (integration of ESSnet results when available: beginning 2012)
Tabular data protection
1
- Recommendation of a model of Confidentiality Charter to be used in different statistical domains: end 2011
- Proof of concept of the new (harmonized) methods for EU tabular data protection (ESSnet project): 2012
- Review the proposal of the redevelopment of Argus into an open source environment and of the governance structure for an open source tool: beginning 2012
- Best practices for Census hypercube protection: a workshop to be organized together with the CENSUS team in Eurostat: beginning 2012
1