Abstract

Statistical agencies should make more use of their high-quality survey data collections to assess the quality of administrative data prior to their use in national statistical systems. In this presentation, I discuss Quality Frameworks for administrative data and propose indicators for assessing the representativeness of administrative data using distance metrics and a representativity (R)-indicator based on the estimation of propensities scores from aggregated auxiliary information available from weighted survey counts. The propensity scores indicate whether baseline covariates are similar between the administrative data and survey data. These propensity scores can then be used as inverse weights to correct representation in the administrative data. I also discuss integrating survey data to administrative data, followed by mass imputation to produce statistical registers. We compare estimates from the statistical register to standard design-weighted and small area estimates from survey data. I conclude with future work in the area of enhancing the use of administrative data in official statistics.