Dr Maximilian Maier, Warwick Business School, University of Warwick

Behavioural interventions—such as reminders to save energy, healthy eating prompts, or opt-out organ donation policies—are increasingly used to address major societal challenges. But a key problem remains: we often don’t know which interventions work best, or why some work in one context but not another. This issue is particularly pressing when trying to synthesize primary literature through meta-analysis, where usually even after accounting for differences between effects based on extant theory a large amount of unexplained variability remains.  

This project aims to develop a new approach to better understand this variation using meta-analytic mixture modelling. This method uses a data-driven approach to identify clusters of interventions with similar effect sizes. Researchers and other stakeholders can leverage these clusters to understand which interventions tend to work best and under what conditions.  

As part of this grant, we will develop these approaches, validate them in a simulation study, create a package that makes it easy for other researchers to use them, and apply them to three large datasets on behaviour change interventions.  

This research will help behavioural scientists, policy makers, and practitioners identify more effective interventions, use evidence more responsibly, and ultimately improve decisions that impact health, sustainability, and well-being. 

Next to development of the R-package, one research paper focused on the substantive insights from applying the mixture modelling to research on nudging will be produced, and one methodological paper showcasing the new mixture modelling technique and the R-package 

A long-term goal is to use the research delivered in this grant as a starting point for a national database of behavioural interventions hosted at Warwick Business School. Once we have established the methodological pipeline, the use of Bayes factor as the inferential tool allows us to update the model sequentially with each new study, resulting in a living systematic review that refines our understanding of heterogeneity as new studies accumulate.