No One-Size-Fits-All: Teaching Quality, Context, and Achievement in TIMSS 2019 and 2023

: (Alternative Format Thesis)

Abstract

Teaching effectiveness is central to improving student learning, yet evidence on how specific classroom practices shape achievement, and for which students, remains mixed. International large-scale assessments (ILSAs) such as TIMSS offer valuable insights by combining achievement data with teacher and student surveys across many countries and cycles. At the same time, their complex design, limited coverage of some constructs, and cross-sectional structure pose challenges for interpretation.

This dissertation investigates the links between teaching quality, captured through instructional clarity, classroom discipline, and cognitive activation, and student achievement in mathematics and science, with attention to how these relationships vary across classrooms and education systems. Four papers address these questions from complementary perspectives using TIMSS data.

Paper 1 maps the architecture of ILSA data and highlights key challenges in analysis and interpretation, including sampling and weighting, plausible values, measurement comparability, and limits to causal claims. It offers a framework for making robust use of TIMSS in comparative education research.

Paper 2 uses a multilevel, variable-centred approach with TIMSS 2019 mathematics to examine whether instructional clarity and classroom discipline are associated with achievement and whether these links differ by students’ socioeconomic status. The results support positive main effects in many systems while showing that equity-relevant interactions are context-dependent.

Paper 3 adopts a person-centred perspective in TIMSS science, using latent class analysis to identify teacher profiles based on general and subject-specific cognitive activation strategies and relating these profiles to achievement. The findings reveal patterned combinations of practice that averages can obscure and show that their effectiveness varies across systems.

Paper 4 introduces a climate-strength lens with TIMSS 2023 mathematics and science, analysing not only mean classroom levels of clarity and discipline but also the dispersion of student perceptions. Mean levels generally align with higher achievement; dispersion often corresponds to lower achievement, though moderation patterns vary by context, indicating that consistency and differentiation can both matter.

Together, the papers show that teaching quality relates to learning through multiple pathways: average levels of practice, differences between student groups, heterogeneity in teacher profiles, and the degree to which students share classroom experiences. Instructional clarity and classroom discipline are broadly beneficial; cognitive activation is valuable but context-sensitive, especially in its subject-specific forms; and no single approach works everywhere. Methodologically, the thesis demonstrates how to use ILSA data responsibly, combining multilevel moderation, person-centred profiling, and climate-strength modelling, to generate results that are comparable, nuanced, and informative for research, policy, and teacher development. As a whole, the thesis offers a transferable template for using ILSA data in teaching effectiveness research.

Date of Award	20 May 2026
Original language	English
Awarding Institution	University of Bath
Supervisor	Andres Sandoval Hernandez (Supervisor) & Andrea Abbas (Supervisor)