Dr ZhiMin Xiao

Causality@Exeter: Seminar Series - Individualised Treatment Effect in Educational Interventions

Open to University of Exeter staff and students

An Institute for Data Science and Artificial Intelligence seminar
Date12 April 2021
Time14:00 to 15:00
PlaceOnline

Join us for the final seminar in this series looking at Causality research at Exeter.

Dr ZhiMin Xiao - Graduate School of Education: Individualised Treatment Effect in Educational Interventions

Understanding whether one thing causes another is a central goal of much of data science.  For example, understanding causal and effect relationships allows us to answer questions such as “Does this treatment harm or help patients?”  However, much of data science, machine learning and statistics is built on correlations.  This seminar series brings together researchers across the University working on and using causal analysis with the aims of understanding different approaches to causal analysis and developing new collaborations and methods.  Starting with University of Exeter experts in causal analysis, the seminars will expand to include external visitors. 

Register to attend the seminar here.

 


Abstract

The use of large-scale Randomised Controlled Trials (RCTs) is fast becoming “the gold standard” of testing the causal effects of policy, social, and educational interventions. RCTs are typically evaluated — and ultimately judged — by the economic, educational, and statistical significance of the Average Treatment Effect (ATE) in the study sample. However, many interventions have heterogeneous treatment effects (HTE) across different individuals, not captured by the ATE. One way to identify HTE is to conduct subgroup analyses, such as focusing on low-income Free School Meal (FSM) pupils as required for projects funded by the Education Endowment Foundation (EEF) in England. These subgroup analyses are usually not standardised across studies and offer flexible degrees of freedom to researchers, potentially leading to mixed results. While the uncertainties in estimates of treatment effects associated with subgroup analyses can be ameliorated by standard or even individual pupil-level data (IPD) meta-analyses, the synthesis of evidence from such methods relies heavily upon the comparability of outcome measures and the consistency in design and implementation across trials. In this talk, I demonstrate how the pooled effects for literacy and maths, through conventional and IPD meta-analyses of 88 large-scale educational interventions, helped decision-makers answer “what worked” for sure, when, where, and for whom. But they also posed new problems that I believe Individualised Treatment Effect (ITE) using machine learning holds the promise to solve. The findings presented in this talk have implications for decision-makers in education, public health, and medical trials.

Add this event to your calendar
Bookmark and Share