News
LISER organises an International Best Practice Workshop on methodology improvements for EU-SILC
A contribution to methodological improvements of EU-SILC: population representativity, imputation, weighting…
Essex, 20-22 February – The “Third Network for the Analysis of EU-SILC” (Net-SILC3), which gathers experts from nine European Statistical System (ESS) bodies and eight academic bodies, organised a key European event on non-response, calibration and weighting. This event was hosted by the University of Essex. It was organised by the Luxembourg Institute of Socio-Economic Research [LISER] which ensures the coordination of Net-SILC3 and brought together participants from the whole Europe.
This workshop presented preliminary Net-SILC3 findings on frame errors and non-response, weighting and calibration and imputation for income variables.
The papers presented at this event are part of a broader analysis of the range of non-sampling errors in the context of EU-SILC data. They will contribute to the development of common tools and approaches regarding various aspects of data production for the whole European Statistical System. Net-SILC3 is funded by the Statistical Office of the European Union (Eurostat). It builds on a solid foundation of previous EU networks coordinated by LISER (esp. Net-SILC1 and Net-SILC2). Launched in 2016 for a period of four years, Net-SILC3 will organise another international conference as well as one additional best practice workshop on the impact of different mode of data collection in 2020.
Find the details of the programme below:
DAY 1: WEDNESDAY 20 FEBRUARY
13:00-13:30
Registration
13:30-14:00
Opening session: Welcome and introduction
Chair: Eric Marlier, LISER/ Net-SILC3 coordination
- Eric Marlier (LISER)
- Emilio Di Meglio (European Commission, Eurostat)
- Peter Lynn (ISER, University of Essex
14:00-15:45
Session 1: Frame errors
Chair: Eric Marlier (LISER)
Frame errors in EU-SILC: Under-coverage & Exclusions from target population – Tara Junes (Statistics Finland)
Presentation 1 | Presentation 2
Q&A
Panel
- Under-coverage/exclusions from target population and the EU “Integrated European Social Statistics” instrument (Emilio Di Meglio, European Commission, Eurostat)
- Under-coverage/exclusions from target population and the quality and definition of EU social indicators (Rudi Van Dam, Social Protection Committee’s Indicators Sub-Group)
- Under-coverage/exclusions from target population from a research point of view (Anne-Catherine Guio, LISER/Net-SILC3 coordination)
Presentation
Overall discussion
15.45-16:15
Break
16:15-17:45
Session 2: Non-response and representativeness
Chair: Gareth James (Statistics UK [ONS])
Impact of panel attrition on representativeness of EU-SILC – Barry Schouten (Statistics Netherlands [CBS])
Comparisons of representativeness of response of the EU-SILC first wave across countries – Barry Schouten (Statistics Netherlands [CBS])
Discussant: Li-Chun Zhang (University of Southampton and Statistics Norway)
Discussion
DAY 2: THURSDAY 21 FEBRUARY
09:00-10:30
Session 3: Non-response and proxy interviews
Chair: Lars Lyberg (Inizio, Inc.)
Effect of proxy responses on non-response error – Peter Lynn (ISER, University of Essex)
Q&A
Best practice on minimising non-response error – Peter Lynn (ISER, University of Essex)
Illustration of best practice – Stephan Häni (Statistics Switzerland)
Presentation
Discussion
10.30-11:00
Break
11.00-12:30
Session 4: Calibration and weighting (1/3)
Chair: Peter Lynn (ISER, University of Essex)
Comparative review of weighting – Martins Liberts (Statistics Latvia [CSB])
Q&A
Use of registers in calibration – Martins Liberts (Statistics Latvia [CSB])
Q&A
Illustration of best practice – Barry Schouten (Statistics Netherlands [CBS]) and Frankovi? Boris (Statistics Slovakia)
Presentation 1 | Presentation 2
Discussion
12.30-14:00
Lunch
14.00-15:15
Session 5: Calibration and weighting (2/3)
Chair: Emilio Di Meglio (European Commission, Eurostat)
Weighting for a modular structure – Andy Fallows (Statistics UK [ONS])
Q&A
Weighting together panels for cross-sectional estimation – Olena Kaminska (ISER, University of Essex)
Presentation | Paper
Discussion
15.15-15:45
Break
15.45-16:45
Session 6: Calibration and weighting (3/3)
Chair: Barbara Moench (European Commission, Eurostat)
All Abstracts | All Papers | Presentation
Best practice for calibration and weighting – Gareth James (Statistics UK [ONS])
Discussion
Wrapping-up (unit non-response and weighting) – Peter Lynn (ISER, University of Essex)
DAY 3: FRIDAY 22 FEBRUARY
09:00-10:45
Session 7: Item non-response and imputation
Chair: Tim Goedemé (University of Antwerp and University of Oxford)
Comparative analysis of item non-response of income variables – Richard Heuberger (Statistics Austria)
Q&A
Imputation practices for income variables – Sophie Psihoda (Statistics Austria)
Q&A
Net-gross conversion procedures:
Comparative analysis and best practices – Richard Heuberger (Statistics Austria)
Discussant: Chrysa Leventi (Council of Economic Advisors of the Greek Ministry of Finance)
Overall Discussion
10.45-11:15
Break
11:15-12:45
Session 8: Imputation: innovative/alternative methods and best practice
Chair: Guillaume Osier (Statistics Luxembourg [STATEC])
Potential for longitudinal imputation – Nadja Lendle (University of Linz)
Q&A
Best practice for imputation – Richard Heuberger (Statistics Austria)
Illustration of best practice – Doriana Frattarola (Statistics Italy [ISTAT])
Overall discussion
Wrapping-up (unit non-response and weighting) – Hans Kiesl (University of Regensburg)
12:45-13:15
Closing session
Chair: Anne-Catherine Guio (LISER/Net-SILC3 coordination)
Workshop conclusions and next steps – Peter Lynn (ISER, University of Essex)
Final words – by Eurostat and Net-SILC3 coordination