Register-based statistics are the basis for the 2011 Census in many countries. We will look at methods for doing quality evaluation of register-based statistics, and we will use the Norwegian register-based employment statistics as a case study. This statistics is an example of micro-integration, where several administrative registers are integrated on unit level. For quality evaluation we will use the Labour Force Survey as a reference or target data, the source being closest to the employment status as defined by ILO, the International Labour Organisation. The register-based employment data are surrogate data for the target data. Traditionally the measure of discrepancy between the surrogate and the target data has been unit misclassification. However, the necessary linking on unit level is impossible in many countries. The measure can also be misleading when the focus is the quality of statistics and not the quality of micro-data. We will compare the mean squared errors of statistics based on the surrogate data with the mean squared errors of the statistics based on target data. The focus will be on small areas.
Keywords: Micro-integration; Register-based statistics; Small area statistics; Data-integration
Biography: Johan Fosen took a Ph.D in 2007 at Department of Biostatistics, University of Oslo, and after that he has been working at Statistics Norway. For the last couple of years, one main area of interest has been quality evaluation of registers and register-based statistics.