A STUDY OF MULTILEVEL ANALYSIS

Loading...
Thumbnail Image
Date
2003-08
Authors
WAH, CHAN AH
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Multilevel analysis is being applied in a growing number of medical and social science research areas such as epidemiology, voter behaviour, sociology, geography, educational and organizational studies. It is getting very popular these days. In this study, we aim to explain the essentials of multilevel analysis, how it is carried out in 1:\vo very popular software packages, and some of the controversial issues researchers have to face when using this method. We hope that this dissertation will enable the reader to gain a more realistic perspective towards this 'new' statistical tool. Researchers from different academic fields use different tem1s or phrases when they deal with multilevel problems. As such, it has been a source of confusion for many people, especially for those who are new to this area. We shall clarify this matter right from the start. We shall also compare multilevel techniques with ordinary least squares method, followed by a brief outline of the historical roots of the subject. Next, we shall look into the mathematical theory behind multilevel analysis, parameter estimation techniques, hypothesis testing, and other standard statistical concepts would also be discussed. Two of the most popular multilevel software packages (HLM and MLwiN) are then tested and their applications explained. We shall also examine some of the fundamental challenges facing multilevel analysis and their possible solutions. Finally, in the Appendix, we have included as reference a glossary for multilevel analysis, a short article on different multilevel terms with the same meanings, and a short survey of major multilevel software packages
Description
Keywords
MULTILEVEL ANALYSIS
Citation