The aim of the study was to investigate causal effects of teacher characteristics and school climate on mathematics achievement, taking advantage of data from TIMSS. International large-scale studies, such as TIMSS, typically have a cross-sectional design. With such designs it is difficult to establish causal relations between variables, because observed relations are influenced by omitted variables, reverse causation, and errors of measurement (Murnane & Willet, 2010). However, by investigating within-country change over time, biasing influence from omitted variables in the form of fixed country characteristics is avoided, thereby increasing the possibility to make correct causal inferences. Furthermore, with aggregated data, errors of measurement have only little impact. The analyses therefore were based on a panel data set formed by data aggregated to the country level from the 38 countries which participated with grade 8 in both TIMSS 2007 and TIMSS 2011 (Mullis et al., 2012). Different methods have been devised in order to control for the effect of the units which are subject to repeated measurements, such as regressions based on difference scores, or regression analyses in which units are identified with dummy variables. Here we applied structural equation modelling (SEM) techniques, which offer several advantages such as use of both latent and manifest variables, and possibilities to compare models based on different assumptions (Bollen & Brand, 2008). The analyses focused teacher characteristics (educational level, teaching experience and major academic discipline studied), professional development, self-efficacy and school climate (SEAS), which in previous research have been shown to influence mathematics achievement (Kyriakides et al., 2010; Wang & Degol, 2015). Results showed that the teachers’ attained level of education had effects on mathematics achievement. Quite substantial effects of professional development on student achievement were also identified. Teacher self-efficacy, as assessed by self-reports of preparedness for teaching in different domains, was positively, but insignificantly, related to student achievement. No effects were found of the teacher characteristics years of teaching experience and the major academic discipline studied. SEAS did not satisfy ideals of unidimensionality, and only items reflecting parental support for student achievement and students’ desire to perform well were significantly related to student achievement. No differences were found in the strength of effects for OECD and non-OECD countries. It is concluded that there was both agreement and disagreement between current and previous findings, and that the reasons for the disagreements need to be investigated in further research.

Bollen, K. A., & Brand, J. E. (2008). Fixed and Random Effects in Panel Data Using Structural Equations Models. California Center for Population Research, On-Line Working Paper Series, PWP-CCPR-2008-003 Kyriakides, L., Creemers, B., Antoniou, P., & Demetriou, D. (2010). A synthesis of studies searching for school factors: implications for theory and research. British Educational Research Journal, 36(5), 807-830. doi: 10.1080/01411920903165603. Mullis, I.V.S., Martin, M.O., Foy, P., & Arora, A. (2012). TIMSS 2011 International Results in Mathematics. Chestnut Hill, MA: TIMSS & PIRLS International Study Center, Boston College. Murnane, R. J., & Willett, J. B. (2010). Methods matter: Improving causal inference in educational and social science research. Oxford University Press. Wang, M.-T., & Degol, J. L. (2015). School Climate: a Review of the Construct, Measurement, and Impact on Student Outcomes. Educational Psychology Review, 1-38.