Options
Analysis of TIMSS data
Author
Foo, Seau Fah
Supervisor
Lau, Shun
Abstract
About 5000 Secondary 2 students from all secondary schools in Singapore, together with their principals and Mathematics and Science teachers, participated in the Third International Mathematics and Science Study (TIMSS) 1999. Extending beyond the methodology used and the TIMSS-related findings reported for Singapore in previous studies, this within-country secondary analysis of Singapore's TIMSS 1999 data employed 2-level hierarchical linear modelling to determine the student-level and classroom-level predictors of Mathematics and Science achievement and to ascertain their relative influences on student achievement.
At the student level, this study found that students' perceptions of their competence in Mathematics/Science, their interest in Mathematics and their beliefs in the efficacy of Science predicted their achievement in these subjects positively, even after controlling for other predictors. This suggests that beyond students' socio-economic status which is less malleable to teachers' influence, teachers can foster students' beliefs that their competence in the subject is changeable and under their control, stimulate students' interest in the subject and show students the connections between Science as taught in the classroom and its real-world applications so as to enhance student achievement in these subjects.
At the classroom level, aggregated socio-economic status was the strongest positive predictor of both Mathematics and Science achievement, accounting for over half of the between-classroom variance in each case. This indicates that in the Singapore context, what classmates a student studies with in his/her class appears to be more important than what family he/she comes from in predicting his/her Mathematics and Science achievement. Teachers' emphasis on classroom practices involving homework in Mathematics/Science and their assignment of extended type of Mathematics homework also predicted achievement in these subjects positively, after controlling for other predictors. This suggests that the way teachers treat homework and the type of homework they assign to students are critical for student achievement.
This study also noted certain features of TIMSS 1999's design which constrain within-country multilevel secondary analyses of its data. To open up a richer array of research possibilities for future TIMSS secondary analyses, future TIMSS studies could select students from more classrooms per school so that 3-level hierarchical linear models examining both classroom and school effects within participating countries can be specified. A longitudinal research design or a pre- and post-test design could also be adopted to increase the power of analysis for teacher or instructional effects. More critically, to improve the quality of scales arising from secondary analyses of TIMSS data, future TIMSS studies could draw on established theoretical frameworks of educational effectiveness to guide the operationalisation of more valid and reliable explanatory factors of student achievement.
At the student level, this study found that students' perceptions of their competence in Mathematics/Science, their interest in Mathematics and their beliefs in the efficacy of Science predicted their achievement in these subjects positively, even after controlling for other predictors. This suggests that beyond students' socio-economic status which is less malleable to teachers' influence, teachers can foster students' beliefs that their competence in the subject is changeable and under their control, stimulate students' interest in the subject and show students the connections between Science as taught in the classroom and its real-world applications so as to enhance student achievement in these subjects.
At the classroom level, aggregated socio-economic status was the strongest positive predictor of both Mathematics and Science achievement, accounting for over half of the between-classroom variance in each case. This indicates that in the Singapore context, what classmates a student studies with in his/her class appears to be more important than what family he/she comes from in predicting his/her Mathematics and Science achievement. Teachers' emphasis on classroom practices involving homework in Mathematics/Science and their assignment of extended type of Mathematics homework also predicted achievement in these subjects positively, after controlling for other predictors. This suggests that the way teachers treat homework and the type of homework they assign to students are critical for student achievement.
This study also noted certain features of TIMSS 1999's design which constrain within-country multilevel secondary analyses of its data. To open up a richer array of research possibilities for future TIMSS secondary analyses, future TIMSS studies could select students from more classrooms per school so that 3-level hierarchical linear models examining both classroom and school effects within participating countries can be specified. A longitudinal research design or a pre- and post-test design could also be adopted to increase the power of analysis for teacher or instructional effects. More critically, to improve the quality of scales arising from secondary analyses of TIMSS data, future TIMSS studies could draw on established theoretical frameworks of educational effectiveness to guide the operationalisation of more valid and reliable explanatory factors of student achievement.
Date Issued
2006
Call Number
QA14.S55 Foo
Date Submitted
2006