Quantitative Methods in Education Masterclass Series (Spring 2026) - Methodological Trade-offs between Machine Learning and Traditional Statistical Models in Complex Survey Data
Date:
05/05/2026
Organised by:
Southampton Education School in collaboration with the National Centre for Research Methods
Presenter:
Dr Yin Wang
Level:
Entry (no or almost no prior knowledge)
Contact:
Jacqui Thorp
Training and Capacity Building Coordinator, National Centre for Research Methods, University of Southampton
Email: jmh6@soton.ac.uk
Venue: Online
Description:
DESCRIPTION
Tuesday 5 May 2026
15:00–16:30 (UK time)
Online (MS Teams): Link to Meeting
ABSTRACT
This session focuses on a central question: when data arise from complex sampling designs, can machine learning methods support rigorous inference in the same way as traditional statistical models? We begin by contrasting the fundamental objectives of the two approaches: while traditional statistical models emphasise parameter estimation and uncertainty quantification, machine learning methods tend to prioritise predictive performance.
The session then considers what this distinction implies in the context of complex survey data. In particular, it focuses on key design features such as sampling weights, clustering structures, and plausible values, which are essential for valid inference but are often not systematically addressed within machine learning frameworks. Using empirical analysis based on TIMSS 2023 data, the session illustrates how different methodological approaches handle these features, and how these choices shape our understanding of how students’ learning behaviours influence learning outcomes.
SHORT BIO:
Dr Yin Wang is a Lecturer in Research Methods and AI Skills in the Department of Social Statistics and Demography at the University of Southampton, and a member of the National Centre for Research Methods (NCRM). Her current work within the UKRI-funded programmes Using Artificial Intelligence Methods in Education Data and New Approaches to Digital Skills Development examines the integration of AI-based and traditional statistical methods to improve the validity, transparency, and policy relevance of quantitative research using international large-scale assessment (ILSA) data.
Cost:
Free of charge and registration is not required - 'Register for this course' below links to the Teams Meeting.
Website and registration:
Region:
South East
Keywords:
Quantitative Data Handling and Data Analysis, Machine learning, Quantitative Software, Sampling Designs, Survey Data
Related publications and presentations from our eprints archive:
Quantitative Data Handling and Data Analysis
Machine learning
Quantitative Software