Table 1.
Five categories of motivations for modelling, illustrative prompt questions, and focus during model construction.
Motivations for modelling | Illustrative prompt questions or statements and references | Focus during model construction | |
---|---|---|---|
Mapping and formalising theory—providing conceptual frameworks | Modelling helps provide a conceptual framework (for self or others) | What are the key entities and processes required to model bTB and how might we formulate them in the most conceptually useful way? | Conceptual clarity of key entities and processes and formalisation of these |
Begin with informal understanding or verbal theory; obtain a precise formal representation of the theory (a full model) or of concepts and subcomponents of it | Are there similar concepts in associated areas that could apply (e.g., how does reproductive potential relate to R0)? (9) | ||
Exploring theory—exploring possibilities | A model formalising theory is used to constrain relationships between entities so that system behaviours can be explored | Is infection invasion success dependent on spatial clustering? (10) | Accurate representation of relevant aspects of the theory in the model |
Begin with a model that formalises theory; obtain a set of possible behaviours given those processes | What is the probability of bTB persisting in cattle herds of different sizes? (11) | No explicit use of data is required | |
Building theory—generating hypotheses and explanations | The structure of the formalised model focuses our attention on particular processes and parameters, changes to which constitute testable hypotheses | Following the 10-year randomised badger culling trial, bTB incidence in cattle decreased in the badger culling area, but increased in adjoining areas (12) | Observing the way structures and parameters suggest model reformulations |
Begin with an observation or data; obtain precise hypotheses. NB: theory building often conducted iteratively with theory testing (below) | The 1967–1968 UK foot-and-mouth disease (FMD) epidemic was characterised by rapid early spread followed by slower later spread (13) | ||
Testing theory—identifying mechanisms | To generate empirically relevant and measurable predictions, for the purpose of falsification | Does the incorporation of transmission heterogeneity allow us to better explain the data? (14) | Incorporation of mechanisms into a model in ways that allow us to establish whether observed phenomena can be reproduced; structural equivalence between data and model outputs |
Begin with a model that encapsulates a theory; obtain predictions that can be compared with data to help pinpoint incorrect mechanisms | |||
Applying theory—generating accurate predictions | To make forecasts, predict responses under intervention, and examine counterfactual scenarios | How might FMD epidemiological dynamics have differed under alternative culling scenarios during the 2001 FMD outbreak? (15) | Ensuring key mechanisms are replicated as closely as required to accurately reproduce real-world phenomena and data |
Begin with a model that is assumed to be true; obtain hindcasts/forecasts, and predictions relating to counterfactuals and other systems | What difference might incursion location and speed of deployment make to the effectiveness of FMD reactive ring vaccination? (16) |