Do findings from a single country generalise internationally? Evidence from an international assessment

International large-scale assessments across dozens of countries offer an unparalleled opportunity to study education systems worldwide. Yet this potential remains largely underused. Despite their global scope, most research based on these data focuses on single countries at a single point in time, limiting generalisability. We examine implications for comparative research and illustrate this using an individual participant data (IPD) meta-analysis. Using data from PIRLS 2011, 2016, and 2021, we use a conventional multilevel model for each participating country and year. We synthesised the resulting coefficients using random-effects meta-analysis. The findings showed differences in generalisability: some predictors exhibit stable associations across countries and time, whereas others vary substantially across contexts. This demonstrates the importance of evaluating the generalisability of effects rather than relying on single-country studies. Overall, the study illustrates how meta-analytic approaches applied to international assessment data can improve the robustness and cumulative value of educational research.