The aim of the modelling seminar was to discuss, with people who use models to solve similar challenges, the value of modelling/analysing complex problems - especially where the behaviour of individuals and the system is highly uncertain.
Having summarised the strengths and weaknesses of different approaches, using examples from some of our recent projects to illustrate different techniques, the meeting went on to debate the issues. A number of themes emerged during the discussion:
- Uncertainty in models needs to be recognised (whether it arises from the underlying data, modelling assumptions or modelling approach) and spurious accuracy in the results must be avoided.
- Conveying the uncertainty of results in ways that help, rather than hinder, decision makers is important (e.g. quantifying the probability of a 'successful' or 'unacceptable' outcome for different options, rather than presenting distributions).
- Integrating established models can help ensure that the effects of feedback from resource constraints, economics, etc. on policy options are considered.
- Top down models are often helpful when deciding on the scope of and approach to more detailed models.
- Models take a long time to develop so it is important to anticipate the types of question that could be asked and ensure that models can be adapted to consider new questions.
- Care must be taken to ensure that models are not applied to problems they were not designed to address. It is therefore important to retain staff who understand the strengths and weaknesses of the models.
- It is often difficult to validate models based on real life events - some of which may be decades ago and are no longer relevant - especially when the model does not match actual experience. In this context, comparing outputs from different models can be valuable.
A follow-up session is now being planned to focus on some of these themes.