Molding TIMSS Data: Botswana Data as an Example

Modeling of TIMSS data: This study will be of great use to many Behavioral and Social researchers including graduate students who use quantitative methods to analyze nested (or multilevel data). A two-level hierarchical linear modeling technique is employed. Missing data issues are discussed. In particular, a comparison is made between the results obtained when treating missing data at level-2 using listwise deletion and by substitution using expected maximization. Briefly,the study offers a holistic examination of mathematics learning in schools taking Botswana as an example using data from IEA s Trends in International Mathematics and Science Study (TIMSS) of 2003. The analyzed sample was comprised of 4,805 students in 136 schools. Part of the theoretical framework for variable selection was based on six factors from the Rand model. The six factors were student background and attitude (at the individual- level) and school-quality, teacher-attribute, and instructional-quality (at the school-level), with mathematics achievement as the outcome factor.

Autorentext

Poloko Chepete:PhD:Studied Curriculum and Instruction major in Mathematics Education and minor in Inquiry Methods at Indiana University. Taught math in schools and Colleges of education in Botswana.Currently head Mathematics Education Department.Research interests:large scale assessments HLM and SEM:recipient Clyde & Lineback Fellowship IU.