Development of Low-stakes Mathematics and Literacy Test Scores during Lower Secondary School – A Multilevel Pattern-Centered Analysis of Student and Classroom Differences

Abstract The development of students’ learning and test-taking behavior may derive from the social context and the group of peers they associate with daily for years. Consequently, it is assumed that students’ academic achievements are to some degree affected by their classmates and the composition of the classroom. The present study provides evidence on how Finnish students (N=5071) from different classrooms (N=435) develop distinct patterns regarding their mathematics and literacy achievement during lower secondary school. We analysed longitudinal large-scale educational assessment data using a multilevel latent profile analysis (MLPA) to investigate the impact of classroom effect on students’ achievement patterns, that is, on the development of students’ low-stakes mathematics and literacy test scores from 7th to 9th grade. The results demonstrated the added value of modelling the multilevel structure inherent in educational assessment data: we identified four student achievement patterns that displayed different distributions across the school classes. More precisely, besides individual characteristics, the development of students’ low-stakes mathematics and literacy test scores was associated with class-level factors and some of the classrooms seemed to have a stronger effect on students’ test scores. These results suggest that classroom context is associated with students’ achievement patterns, especially regarding the worst achieving students. The findings may reflect a combination of class placement practices as well as classroom and peer effect. Although the differences between Finnish schools have been one of the lowest in the OECD countries, the findings of the present study suggest that the classroom membership may create class level quality differences in both the preconditions and the development of learning.