Factors enclosed in the concept Opportunity to Learn (OTL) are crucial in the context of international studies where educational systems are compared, not only for interpreting the variation in students' test results, but also for evaluating the quality and efficiency of the educational environments. The degree to which the tasks included in the measurement instruments represent the educational goals and the actual teaching in the different systems will be investigated.
The OTL concept
One aspect of the OTL concept represents the degree to which the content of the tests is aligned with curricula and actual teaching content. While OTL-information has been systematically collected from teachers and through analyses of curricula and text-books, few attempts have been made to connect this information to analyses of the achievement data. The most ambitious OTL analyses were conducted in relation to the TIMSS 1995 study (Schmidt et al., 1997). One of the reasons for the limited use of the OTL information is the analytical challenge inherent in investigating data collected at differing levels of measurement.
Within the sub-project analyses will be made in which achievement data is related to the OTL information. One approach that will be used is to reduce the large number of OTL items to a more manageable set of latent variables, and then to use two-level SEM to relate these to achievement data at class-level.
A special focus in this sub-project is on the allocation of time, both in absolute and in relative terms. In the OTL framework, time is a theoretically important concept relevant to educational policy. Deficiencies in earlier methodological approaches have implied that direct effects of amount of time allocated to mathematics instruction to achievements are hard to discern. However, studies utilizing more robust analysis suggest that time allocation is an important causal factor, and viable approaches for investigating these effects are available. Analyses will be conducted to capture interaction effects with individual-level variables using regression models with country-level fixed effects (Angrist & Pischke, 2009).