Seidel and Shavelson (2007) observed, in a meta-analysis of research on effects of teaching, that categorization of teaching variables differs considerably between studies which makes it difficult to integrate findings. Nevertheless, this and other reviews (e.g., Hattie, 2009; Walshaw & Anthony, 2008) indicate that it is essential that teaching is cognitively challenging and provides well-structured learning opportunities; that it offers learning support through monitoring of the learning process, individual feedback, and adaptive instruction; and that classroom- and time management is efficient (cf. Baumert et al, 2010).
While PISA data offers limited information about teaching, the student- and the teacher-questionnaires administered in the IEA studies provide self-reported information, by teachers and students, about the way the teaching is organized and conducted within the classrooms. Information about different aspects of the school, such as school climate is also provided through the school leader’s questionnaire. Our own previous research indicates that by deploying structural equation modelling which draws upon the possibility to separate student level information from class- and school-level information, and to represent the information with error-free latent variables, it is possible to identify a variety of classroom and school factors affecting educational outcomes. This will be a major methodological approach in this project. The major threat to correct causal inferences in this approach is to adequately control for effects of levels of previous knowledge, given the cross-sectional design. However, the relatively powerful control possibilities within the two-level SEM approach allow systematic analyses of unequal distribution of teaching quality over social groups. Analysis of fixed student effects over different subject-matter areas (e.g., language versus science) is another powerful approach to impose such control.
Consistent and interpretable results
One of the reasons why research on teacher characteristics has met difficulties reaching consistent and interpretable results is that it has typically not investigated characteristics of teaching as mediators between teacher characteristics and student outcomes. Baumert et al. (2010) demonstrated the power of such an approach, which will be used in the sub-project. It thus is of particular importance to take seriously the composition of the student group and the various mechanisms that influence the study environment teacher-student interaction. And “the key question is how these effects are mediated, and in particular whether contextual influences operate primarily through peer processes, or also impact on school functioning” (Rutter & Maughan, 2002, p. 464).
Equity aspects of OTL
As was observed in the review of the research field above it also is essential to explicitly bring in equity aspects of OTL, such as distribution of resources across schools, along with tracking and peer effects, as these affect both level and equity of outcomes (Yang Hansen, Rosén & Gustafsson, 2014). Such analyses will thus be an important part of this sub-project.