An assessment of survey errors in EU-SILC

This Working Paper provides a broad assessment of data accuracy and survey errors in EU-SILC, taking comparability of the results across the national surveys as a basic requirement. A typology of errors in survey data is proposed, distinguishing various categories of errors in measurement on the one hand, and errors arising in the process of estimation from the survey sample to the target population on the other. Important sources and types of errors - including conceptual, measurement, and unit and item non-response errors - are explained and illustrated in empirical terms using EU-SILC.microdata. Examples are provided of the impact on comparability of national differences in data collection and data treatment procedures. Methodology of, and practical problems in, computing sampling errors and design effects for the complex statistics on income and living conditions which can be constructed from EU-SILC are explained and illustrated in detail. The paper concludes with recommendations on improving the potential for the assessment by researchers and other users of data quality in EU-SILC.