When we talk about behavioural interventions, we often need to simplify complex concepts such as ‘context’. And we might reference nudges without fully explaining what they entail in practice. But such simplification can hinder understanding and limit the ability to learn from tested interventions. It leaves critical questions unanswered: What kind of nudge? What does it look like? Who implements it, and when? Who is it targeting? What specific behaviour is it trying to change? Malte Dewies argues that providing a clear and shared terminology to effectively address these questions might help bridge disciplinary divides, foster better collaboration, and enable practitioners to design, implement, and evaluate interventions more systematically and successfully.
We need a comprehensive framework to talk about behaviourally-informed interventions
While behavioural scientists have long recognised that context matters, discussions about nudges are somewhat fragmented. Different disciplines emphasise different aspects of interventions, leading to a lack of cohesion. For instance, policy scholars tend to focus on how nudges shift policymaking, contrasting them with traditional policy instruments. Psychologists often prioritise the psychological mechanisms that make nudges effective. Political economists explore nudges in the context of broader debates about state intervention and individual autonomy. This divergence in focus creates misunderstandings and missed opportunities for cross-disciplinary collaboration, underscoring the need for a more unified and systematic approach.
Recognising these challenges, my co-author Lucia Reisch and I felt the need to propose a classification or perspective that would encourage users to think about nudges more comprehensively. Such a framework would need to acknowledge the nuances, varying perspectives, and the messy reality of changing contexts.
We developed META BI: a tool designed to facilitate a richer and more systematic understanding of nudges. We knew that any meaningful classification system had to reflect existing knowledge while integrating diverse perspectives. To ensure this, we used the Delphi Approach, a structured feedback process aimed at achieving consensus among experts.
We interviewed over 40 experts from academia and the practice of behavioural interventions, gathering their insights to refine our initial classification. This process drew on our understanding of existing theories and evidence in the field, ultimately shaping META BI into a robust and versatile framework.
Introducing META BI: A tool for describing behavioural interventions
Fig 1: The META BI framework
Source: Dewies and Reisch (2025)
META BI stands for Mapping of Environment, Target group, and Agent for Behavioural Interventions. At its core, META BI identifies five key elements of any behavioural intervention:
- Intervention: The behavioural tool or strategy itself.
- Agent: The entity launching the intervention.
- Target Group: The audience at whom the intervention is aimed.
- Behaviour: The specific behaviour the intervention seeks to change.
- Environment: The broader structural and systemic context in which the intervention operates.
For each of these elements, we identified and validated between three and five dimensions as key characteristics of nudge interventions. For example, the ‘intervention’ element is characterised by multiple aspects, including its mechanisms, the psychological principles it leverages. We identified and validated 17 non-exclusive psychological mechanisms that nudges can rely on, such as framing, endorsement, and commitment.
However, mechanisms are just one facet of an intervention. Other dimensions include the format of the intervention (its physical appearance) and its level of intrusiveness, among others. In total, META BI contains 20 such dimensions along which interventions can be described and understood.
Practical applications of META BI
META BI is not just a theoretical framework – it is a practical tool for policymakers, researchers, and practitioners. Users can apply the classification rules and labels we developed for each dimension to think more systematically about the nudges they design and implement. For instance, META BI can help:
- Understand nudging interventions: META BI’s 20 dimensions help users think systematically about interventions, avoiding common pitfalls like ignoring context or unintended effects.
- Enable targeted evidence retrieval: By coding interventions across key dimensions, users might filter and compare behavioural evidence based on mechanisms, agent goals, or target group characteristics, for example.
- Facilitate field mapping and meta-analysis: META BI may enable structured synthesis of behavioural science by capturing factors crucial for generalisability, aiding literature reviews, and explaining heterogeneity in findings.
Bridging disciplines and encouraging dialogue
Ultimately, META BI is more than just a classification system. We view it is a bridge between disciplines and perspectives. By providing a shared language and comprehensive framework, we hope to foster more productive conversations about nudges and behavioural interventions.
For more, read the full article here.
Malte Dewies is a Research Associate at the El-Erian Institute of Behavioural Economics and Policy at the University of Cambridge. He has a background in psychology and extensive experience in public sector engagement. His research is solution-focused and focuses on the intersection of behavioural science and policy to improve policy outcomes.