Direction not Destination

Sunday 3 June 2007

Equifinality

One of the problems of determining the appropriateness of model structure is caused by the presence of equifinality. In order to model open, middle-numbered systems, boundaries need to be drawn on the system to delineate what will be considered in the model and what will not. This positioning of model boundaries, dictating what processes will be represented at which spatial and temporal scales, is known as model 'closure'.

Model closure is not a problem for metaphor models described previously, as the very formulation of those model systems of study ensures they are closed (i.e. they are logically self-contained). But the systems examined and modelled by geographers, ecologists and environmental scientists are inherently open and at scales on the order of the human observer - model closure of the these systems has been in an important point of discussion in these disciplines.

Equifinality is the characteristic of all open systems that a final system state may be reached from multiple initial conditions and via different sequences of system state. In modelling terms, equifinality implies that there are multiple (closed) model structures that may adequately reproduce empirically observed behaviour of an open system. Choosing between these two models then becomes a matter of judgement based on an analysis of the process of model construction - How was the model constructed? What variables were included/excluded? Why? Why not?. Alternatively, the two models might be used in tandem to reflect on what the assumption of each implies for the other and to highlight deficiencies in system understanding.

Thus, equifinality generates uncertainty in the appropriateness of model structure and emphasises that evaluation of the modelling process is as important as evaluation of the model itself.

Labels:

Creative Commons License
This work by James D.A. Millington is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 United States License.

0 Comments:

Post a Comment

<< Home

Read the archives by month:
Read the archives by topic: