02 / Modelling
Making system behaviour visible.
We develop analytical and computational models to investigate how complex social, economic and organisational systems behave. Modelling makes assumptions explicit, connects evidence with possible explanations and provides a way to explore choices before acting.
What connects?
Explore how people, institutions and their environment influence one another.
What could change?
Examine how different assumptions, behaviours or interventions might shape outcomes.
What holds up?
Investigate which findings remain useful when conditions or assumptions change.
Purpose before technique
The question determines the model.
We begin with the system to be understood and the decision the analysis will inform. We examine its boundaries, available evidence, relevant behaviours and sources of uncertainty before selecting a modelling approach.
Our researchers draw on methods they have explored through rigorous scientific work, including soft and hard computing techniques. We assess their suitability for the problem, their data requirements, how their assumptions can be examined and whether their outputs can be interpreted by the intended users.
We often combine complementary methods in a hybrid approach. Each component needs a clear purpose, and the connections between components need scrutiny of their own. The appropriate level of complexity is the one the question and evidence justify.
An iterative process
From a shared question to usable insight.
The work develops through dialogue, testing and revision. Findings may lead us to revisit the framing, the evidence or the model itself.
Frame
Agree the purpose with decision-makers, identify relevant perspectives and define what the model needs to represent.
Build
Select methods, formalise assumptions and connect the model to data and contextual knowledge where available.
Challenge
Check implementation, assess fitness for purpose and investigate uncertainty, sensitivity and alternative explanations.
Explore
Compare scenarios, discuss trade-offs and interpret findings with the people who will use them.
Models open to scrutiny
Confidence comes from careful examination.
We plan checks in proportion to the intended use and the consequences of error. A model can be useful without being a complete representation of reality; its limits must remain visible.
Verification and validation
Verification examines whether the model has been implemented as intended. Validation examines whether it is adequate for its stated purpose, using relevant evidence and review. Neither establishes that every use or conclusion is justified.
Uncertainty and sensitivity
We investigate how assumptions, parameters, data and model structure affect results. Where evidence is limited, we identify the gap and its implications rather than treating precision in an output as confidence in a conclusion.
Documentation and interpretation
We document key assumptions, inputs, methods and checks so that results can be understood and challenged. Scenarios describe outcomes under specified conditions; they do not become predictions simply because they are generated by a model.
Forthcoming demonstrators
Explore the methods in practice.
Selected demonstrators, including AIS and DIS, will be introduced here as they become ready for public use. Each release will explain its purpose, assumptions, development status and limitations.
Public demonstrations will be distinguished from tools assessed for a specific operational use. Exploratory results should be read in the context of their documentation.
- Purpose
- Assumptions
- Development status
- Limitations
Connect the question to the method
What would you like to understand?
Our research develops the scientific foundations. Our advisory work connects modelling with the decisions facing policymakers and organisations.
Discuss a modelling question