Tuesday 14th July      Wednesday 15th July      Thursday 16th July      Friday 17th July     

Thursday 16th July, 09:00 - 10:30 Room: HT-103

Linking survey data and auxiliary data sources: statistical aspects and substantive applications 1
Convenor	Ms Chiara Peroni (STATEC )
Coordinator 1	Mr Francesco Sarracino (STATEC, HSE-LCSR)
Coordinator 2	Mr Wladimir Raymond (STATEC)

Session Details

This session aims to collect contributions from applied research linking survey data with auxiliary data sources. This involves merging various surveys or using information from data collected at different levels, such as macro and individual-level data, administrative data, and other surveys, possibly including those from mobile devices. This permits to address complex research questions, while avoiding the need of long surveys which are costly to run, respond and administer. This strategy, however, poses various methodological challenges concerning the weighting procedure, the computation of standard errors, and the imputation of missing data. These challenges have to be correctly identified and addressed to reach methodologically sound conclusions. We welcome applications to the following topics:

• innovation & social mobility;
• entrepreneurship;
• job and life satisfaction and economic performance;
• migration;

as well as those dealing with methodological issues such as weighting schemes, imputation of missing data and computation of standard errors.

Paper Details

1. Challenges and opportunities of making use of income registers in the EU-SILC weighting procedure
Mr Thomas Glaser (Statistics Austria)

EU-SILC in Austria is a voluntary sample survey on income and living conditions. From EU-SILC 2012 onwards income registers have been used for collecting information on most components of the household income by linking administrative data to the households in the sample. The presence of more comprehensive administrative data in the EU-SILC sample also makes it possible to use this information in course of the weighting procedure. Changes to the weighting procedure due to administrative data - especially in terms of calibration - as well as a unit nonresponse analysis based on income register data will be presented.

2. Weighting schemes for linking data from various survey sources
Professor Takis Merkouris (Athens University of Economics and Business)

Information from different survey sources can be used as auxiliary to each other to produce composite estimates of increased accuracy. Such auxiliary survey sources may be other surveys, subsamples of a single survey, past data from a repeated survey, or administrative data. Possibilities for linking survey data for improved estimation are on the increase in contemporary survey practice. For each case of linked survey micro-data, incorporation of auxiliary information into the sample weighting structure can be accomplished by a suitable weight calibration scheme, which is equivalent to a regression procedure based on the principle of best linear unbiased estimation.

3. Linking survey and aggregated data with crossnational surveys
Miss Rennie Lee (UCLA)

This paper addresses how to link survey data with auxiliary sources. Currently, no data set incorporates individual, coethnic community, and group level factors. I construct coethnic community and group variables with different data sources and append the variables to individuals in nationally representative surveys for the US, Canada, and the UK. My approach for obtaining data is similar in all three countries: a.) individual data is retrieved from nationally representative surveys; b.) coethnic community data is created at small geographic areas using aggregated survey or Census data; and c.) group characteristics are coded from public sources (e.g., World Bank,

4. Using administrative data about employment status to improve the Austrian LFS estimation
Mr Alexander Kowarik (Statistics Austria)
Ms Angelika Meraner (Statistics Austria)
Dr Daniela Gumprecht (Statistics Austria)

The Austrian Microcensus, which incorporates the Austrian Labour Force Survey, is one of the most important sources for labour market statistics. Although the non-response rate is only 5.7 % (in 2013), this non-response results in a small bias for the employment status. By including the employment status from administrative sources into the weighting scheme of the Austrian Microcensus, the bias was reduced.