Reading performance, learning strategies, gender and school language as related issues – PISA 2009 findings in Finland and Estonia.

Reading is considered an important skill not only for academic success, but also for active participation in society. International student literacy assessments report gender differences in reading performance in favour of girls. These reports also show that students from schools with a minority or majority language tend to perform differently: in PISA 2009 in well-performing Finland, the Swedish-speaking schools performed at a lower level compared to the Finnish-speaking schools; in Estonia, the Estonian-speaking schools outperformed the Russian-speaking schools, despite the tests having been translated into each language. How students learn is closely related to their results. In literacy, the more advanced thinking and learning skills known as metacognition enhance the results. Metacognitive awareness can be developed through instruction in the classroom, and this has also resulted in significant improvements for students with rather low learning abilities. As it is teachers’ and schools’ opportunity to help their students by teaching these skills, their awareness of useful strategies could presumably be dependent on the school. So far only the PISA 2009 study has included student awareness of different learning strategies; therefore, the data here enable us to analyse how learning strategies relate to reading, gender or school language. In the current paper, the issues of reading proficiency, learning strategies, gender and school language are considered jointly. Alongside the theoretical background, results from several analyses of PISA 2009 are discussed to show how student awareness and choice of different learning strategies could explain the variation in reading results in boys and girls at student and school levels, and predict their reading test results. The two-level modelling analysis was used as a research method, since it allows us to draw reasonable statistical inferences for regression-type analyses under a hierarchical data structure, and where the factor of individuals being influenced by the group they belong to is explicitly taken into account.