Research project TIMSS (Trends in International Mathematics and Science Study) and PISA (Programme for International Student Assessment) are international comparative studies which aim to measure students achievements. This project aims to examine the relevance in the analytic tools and the statistical models used to analyze TIMSS and PISA in Sweden today.
In the Swedish result report, negligible attention is given to the tools used to generate the results and most of the result presentation consists of simple statistical analysis. The hierarchical data structure (i.e. school, class, and students) and sample procedure in TIMSS and PISA are similar and include giving sample weights to each school, class and student. This project aims to examine the relevance in the analytic tools and the statistical models used to analyze TIMSS and PISA in Sweden today. A further aim is to build multilevel models for the TIMSS and PISA data which takes into account the hierarchical data structure and the sample weights in order to get a more informative and efficient measure of students performance. The modeling phase will be based on statistical theories and the results will be discussed from an educational perspective and compared with other countries as well as examined for trends over time. The planned research is important in an international perspective since it will provide useful insights in how to model international comparative educational studies. In a national perspective it is important to explain Swedish students’ performance so teachers and policy makers can improve the education in Sweden.