The doctoral thesis scrutinizes the content dimension of the Trends in International Mathematics and Science Study (TIMSS) in order to explore two overarching questions: “What content-specific profiles can we obtain about the students’ strengths and weaknesses, and teachers’ instruction, by disaggregating TIMSS into its items and responses?” and “Why is the relationship between science achievement and teachers’ content coverage seemingly so weak in TIMSS?” The thesis consists of four papers (Part II) and an extended abstract that discusses overarching issues (Part I). Paper 1 explores the Norwegian student population’s strengths and weaknesses across the domains and topics in the TIMSS 2011 science test, compared to other content groups and to international averages. Paper 2 investigates patterns in the Norwegian science teachers’ coverage of the TIMSS 2015 content in grades 8 and 9. Paper 3 examines the sensitivity of TIMSS 2015 country rankings in science achievement to differences in class-level content coverage. Paper 4 investigates the sensitivity of TIMSS 2015 science items to Norwegian science teachers’ content coverage in grades 8 and 9. Statistical analyses in Papers 1, 2 and 4 use explanatory item response modelling in a generalized linear mixed modelling framework. General findings from the studies suggest variation in the achievement of science topics (Paper 1) and schools’ content coverage patterns (Paper 2). Despite the variation in achievement and content coverage, the TIMSS test is rather insensitive to variation in content coverage (within a grade) within and between countries when using subject-aggregate measures of achievement and content coverage (Paper 3). Only response-level approaches detect sensitivity (Paper 4), suggesting that a finer-grained analysis is required to pick up on instruction when assessments are remote from the instruction. The thesis also discusses possible issues with the TIMSS content coverage indicators. This thesis belongs to the field of quantitative analyses of international large-scale assessments. The work was carried out at the Centre for Educational Measurement (CEMO), under the Faculty of Education at the University of Oslo.
.