The Research and Development (R&D) initiative has awarded seven projects for its call launched in 2024. In total, R&D has awarded 23 projects across R&D’s four calls, harnessing its aim to advance research and innovation by actively supporting and promoting research projects and activities.
Call 4 was left thematically open, providing an opportunity to highlight a variety of topics within education research that are related to data from IEA studies.
The seven awardees, summarized below, will work on the projects for six months.
An AI-Driven Approach to TIMSS Item Verification and Alignment by Ummugul Bezirhan and Matthias von Davier
TIMSS requires a rigorous item development process to ensure validity, fairness, and alignment with frameworks. The research proposes an AI-driven classification model to streamline development, support the alignment of items with frameworks, and enhance efficiency in addressing challenges such as test assembly, adaptivity, and increasing item demands while supporting the global comparability of items.
SMART: Survey Mode and Response Trends in TIMSS and PIRLS Context Questionnaires by Yuan-Ling Liaw (Linda), Alec Kennedy, Rune Müller Kristensen, and Rolf Strietholt
The shift from paper-based to computer-based surveys in international studies improves efficiency but may reduce participation, risking data quality. We address the challenges of missing data in educational research by examining how survey mode affects response and participation rates in TIMSS and PIRLS and whether impacts vary by student characteristics.
Striving for More Inclusiveness: Understanding and Addressing Student Exclusions in TIMSS and PIRLS by Sabine Meinck, Umut Atasever, Matthias von Davier, Rolf Strietholt, and Bethany Fishbein
Rising exclusion rates pose significant challenges to the validity, representativeness, and cross-national comparability of International Large-scale Assessments. This project analyzes TIMSS and PIRLS data to assess the performance of excluded students and evaluate the impact of their inclusion, providing evidence-based recommendations to improve inclusiveness and strengthen assessment validity, reliability, and international comparability.
Improving IEA Questionnaires for Use in Low- and Middle-Income Countries by Oliver Neuschmidt, Clara Beyer, Sarah Howie, and Katherine Reynolds
This research proposal aims to enhance the relevance and validity of IEA context questionnaires for Low- and Middle-Income Countries. By evaluating regional assessments and initiatives, the project will identify key measures to refine questionnaires, ultimately improving the quality of data to support more effective educational policy interventions.
Information Retrieval Using Retrieval-Augmented Generation on PIRLS Documents by Widianto Persadha and Heiko Sibberns
Using an open-source large language model with Retrieval-Augmented Generation methodology, the research develops a local information-retrieval system. It addresses issues like outdated knowledge, limited domain expertise, and privacy concerns by augmenting it with external information. The outcome is a prototype search engine enabling users to access PIRLS-related information accurately.
Assessment Design for Accommodating Participants with Tailored Support (ADAPT), by Rolf Strietholt, Nurullah Eryilmaz, Julian Fraillon, Olga Kunina-Habenicht, Ana María Mejía-Rodríguez
The ADAPT project explores how countries manage accommodations for students with special education needs (SEN) in IEA studies. By defining SEN categories, analyzing inclusion/exclusion practices, and mapping accommodations, the project aims to inform technical guidelines and promote inclusive assessments, aligning with Sustainable Development Goal 4 for equitable education.
A Two-Step Imputation Approach Combining IRT and Deep-Learning Methods in Large-Scale Survey Assessments by Peter van Rijn, Usama Ali, and Priyadarshini Dwivedi
The research project combines item-response-theory (IRT) and deep-learning methods to impute item responses missing by design. Generative adversarial networks are employed to enhance IRT-based imputations as an alternative to the current plausible-value methodology. The method is to be tested using IEA data and compared with official results to evaluate its viability for reporting.
Are you interested in submitting your project and advancing research and innovation? Call 5 will be announced at the end of June 2025.
All published outcomes from previous R&D calls are available to read open-access on the R&D page.