CSB Working Paper, 2010, 9: The standard error of estimates based on EU-SILC. An exploration through the Europe 2020 poverty indicators

Currently, the European Union Statistics on Income and Living Conditions (EU-SILC) is the single most important data source for cross-national comparative research on income and living conditions in the European Union. As EU-SILC consists of a sample of European households, point estimates should be accompanied by appropriate standard errors and confidence intervals. This is especially so if indicators are constructed for measuring progress towards pre-defined targets such as those of the Europe…

Currently, the European Union Statistics on Income and Living Conditions (EU-SILC) is the single most important data source for cross-national comparative research on income and living conditions in the European Union. As EU-SILC consists of a sample of European households, point estimates should be accompanied by appropriate standard errors and confidence intervals. This is especially so if indicators are constructed for measuring progress towards pre-defined targets such as those of the Europe 2020 strategy. All too often this has been neglected in European poverty research and official publications. In contrast, this paper pays explicit attention to the calculation of standard errors and confidence intervals. Standard errors are strongly dependent on the sample design. Therefore, accurate information on the sample design is crucial, especially for a database like EU-SILC which contains data on about 30 European countries which employ different complex sample designs. However, information on the sample design is incomplete in the EU-SILC User Database for data confidentiality reasons and there are several options for handling this lack of information. In this paper, we document the sample designs used in EU-SILC and compare the information available through different sources, namely the Quality Reports, the User Database and a specific dataset containing additional information about the sample design prepared by Eurostat. Furthermore, on the basis of the specific dataset prepared by Eurostat, we explore which variables are best used when analysing EU-SILC for adequately computing standard errors. We illustrate the importance of various assumptions with regard to the sample design by presenting results for the official Europe 2020 poverty indicators. It is shown that neglecting the sample design can lead to a serious underestimation of the standard errors. In addition, it is discussed how researchers using EU-SILC could best take account of the sample design for appropriately estimating standard errors.