Consistent Estimates for Categorical Data based on a Mix of Administrative Data Sources and Surveys
National Statistical Institutes (NSIs) often use large datasets to estimate population tables on many different aspects of society. A way to create these rich datasets is by utilizing already available register data and supplement them with survey data. A major challenge with the use of combined datasets is to obtain consistent population estimates. Therefore, the main goal of this project is to develop an approach for combining different data sources as effectively and efficiently as possible that can easily be implemented and applied in practice.
In the first project, the focus is on constructing a general class of imputation models that can be used to model the truth, the variable that gives the true value of the conceptual phenomenon that one aims to measure, by making use of multiple indicators within a combined dataset. In the second project, more attention is paid on the relation of the conceptual phenomenon with other variables. The third project focusses on using the models to impute other missing values within the combined dataset as well. Other projects focus on comparing the imputation models with other methods for obtaining consistent estimates within combined datasets and on extensions for longitudinal data.
Prof. J.K. Vermunt, Prof. A.G. de Waal & Dr D.L. Oberski
Tilburg University and Statistics Netherlands
1 March 2015 – 28 February 2019