Investigation of sampling weights in hierarchical regression using PISA sampling structures
Cooperation members:
- Julia Mang (TU München, TUM School of Education, Centre for International Student Assessment (ZIB), Munich)
- Prof. Dr. Helmut Küchenhoff➚ (Ludwig-Maximilians-Universität München (LMU), Departement of Statistics, Munich)
- Dr. Sabine Meinck➚ (International Association for the Evaluation of Educational Achievement (IEA), Hamburg)
- Prof. Dr. Manfred Prenzel➚ (Universität Wien, Centre for Teacher Education, Vienna)
This cooperative project between the Centre for International Student Assessment (ZIB, Julia Mang), the Departement of Statistics at the Ludwig-Maximilan-University (LMU, Prof. Dr. Helmut Küchenhoff), the International Association for the Evaluation of Educational Achievement (IEA, Hamburg, Dr. Sabine Meinck) and the Centre for Teacher Education (University Vienna, Prof. Dr. Manfred Prenzel) focuses on the analysis of different weighting approaches in hierarchical models.
The standard methods for analyzing data and their weightings from large scale assessments cannot be adopted for hierarchical models without adaptation. A question to be answered by hierarchical models would be, for example to what extent students' performance is influenced by school characteristics. Besides a controversial scientific discussion, there is the problem that different software programs use different estimation methods, which lead to different results.
Based on a simulation study, the estimation procedure with appropriate sample weights is worked out that has the least deviation from the actual population characteristics. The simulation is based on the characteristics of the Programme for International Student Assessment (PISA).
The aim is to derive evidence-based recommendations for the methodologically correct performance of multi-level analyses using data from large scale assessments.