|Title||Hierarchical linear models with an application to educational scores.|
|Institute||Thesis (M.Sc.) National Institute of Education, Nanyang Technological University|
|Supervisor||Phang, Rosalind Lay Ping|
|Call no.||HA29 Fan|
In social research work, the structure of the data are often hierarchical. Hierarchical linear models (HLM) is a statistical method that takes this hierarchical structure into account. Hierarchical linear models is a generalization of traditional regression methods. It develops an improved estimation of effects within individual units, the formulation and testing of hypotheses about cross-level effects, and the partitioning of variance and covariance components among levels.
In recent years, it was said that the use of hierarchical linear models in statistical applications, especially for educational data, is a promising but an under-utilized approach. Since hierarchical linear models is developed during this decade, many social researchers are involved in the use of hierarchical linear models and looking for the model testing tools.
The main objective of our research is to use the theory of hierarchical linear models in educational data analysis, comparing it to normal linear regression model and finding out whether hierarchical linear models is a better method for analysing hierarchically nested data. Secondly we created a SPSS matrix language codes which provides Bayes Estimates of random effect and estimates of variance-covariance matrix of Level-2.