Measuring change with the Belgian Survey on Income and Living Conditions (SILC): taking account of the sampling variance

TitleMeasuring change with the Belgian Survey on Income and Living Conditions (SILC): taking account of the sampling variance
Publication TypeOnderzoeksrapport
AuthorsGoedemé, T.
Other Numbers2013/BOEP/UA-EU2020
PublisherHerman Deleeck Centre for Social Policy (University of Antwerp)
Place PublishedAntwerp
Year of Publication2013
Pagination19p.
Abstract

Early 2013 Eurostat published a paper which contains some estimations for the standard error of recent changes in the Europe 2020 poverty reduction indicator (Osier et al., 2013). For Belgium, the published standard errors were contested by national experts. At the same time, Osier et al. (2013) warned for limitations to the accuracy and consistency of the sample design variables, which could be the cause of some counterintuitive results. In this report, I review the Belgian SILC sample design and describe the available sample design variables. In addition, I document the importance of a consistent coding of the sample design variables on the basis of a dataset with consistent sample design variables prepared by Statistics Belgium. In order to provide some idea of the size of the standard error of changes over time, this dataset is used to report a number of estimations of the standard error of recent changes in the at-risk-of-poverty indicator for several vulnerable groups, both at the national and regional level. It is concluded that the power of the Belgian SILC data is not sufficient for closely monitoring the situation of relatively small vulnerable groups, especially at the regional level. Apart from investing more resources in the Belgian SILC data, it is recommended that consistent sample design variables are included in all available SILC datasets and that further research is carried out about the most appropriate estimator of the sampling variance for the Belgian SILC data. In the meantime, the use of the ultimate cluster method for estimating the sampling variance in European research and by Eurostat, appears to be a pragmatic but valid solution, provided the sample design variables are coded adequately.

KeywordsBelgium, EU-SILC, Europe 2020 poverty reduction target, sample design, sampling variance, Standard error
Citation Key4081
Universiteit Antwerpen - University of Antwerp
Centrum voor Sociaal Beleid © 2005 - 2019