School emphasis on academic success: exploring changes in science performance in Norway between 2007 and 2011 employing two-level SEM

Abstract

We study whether changes in school emphasis on academic success (SEAS) and safe schools (SAFE) may explain the increased science performance in Norway between TIMSS 2007 and 2011. Two-level structural equation modelling (SEM) of merged TIMSS data was used to investigate whether changes in levels of SEAS and SAFE mediate the changes in science performance. Two mediation models were fitted, one using subdomain scores of science as manifest dependent variables and one in which these scores are indicators of a latent science performance variable. The change in the latent science variable was fully mediated by SEAS, but this model did not explain changes in earth science performance, which increased more than the other subdomains. In the model with subdomain scores as manifest dependent variables, SEAS mediated the increased performance of all 4 subdomains of science. SAFE did not explain increased science performance but did have a positive impact on SEAS.

Keywords:

• school emphasis on academic success
• safe schools
• science
• mediation models
• SEM

Notes on contributors

Dr Trude Nilsen is working for the Norwegian TIMSS project at the University of Oslo. She has also been working with professional development of teachers at the Physics Department of Oslo, and before that she was a high school teacher in physics and mathematics for 10 years. Her research focuses on science education with a special interest in large-scale international studies and quantitative methodology.

Jan-Eric Gustafsson has since 1986 been Professor of Education at the University of Gothenburg. His substantively oriented research focuses on individual differences in cognitive abilities, and on determinants of educational outcomes at individual and system levels. His methodologically oriented research focuses on conceptual and technical issues of measurement within classical and modern test theory, and on the application of latent variable models to both non-nested and nested data.