Gedetailleerde leidraad

A Appendix: Uncertainty Matrix

The uncertainty matrix is an aid in making an inventory of where ('locations') the most (policy) relevant uncertainties are expected, and how they can be characterized in terms of a number of uncertainty features. It can serve as a first step of a more elaborate uncertainty assessment, where the size of uncertainties and their impact on the policy-relevant conclusions are explicitly assessed. The matrix [2] is structured in five principal dimensions, 'location', 'uncertainty level', 'nature of uncertainty', 'qualification of knowledge base', 'value-ladenness of choices', which are further explained below:

[2] For this appendix we have made extensive use of the material presented in the recent paper of Walker et al. 2003. In that paper a typology and an associated uncertainty matrix was presented which classify uncertainty according to three dimensions: its 'location' (where it occurs), its 'level' ( where uncertainty manifests itself on the gradual spectrum between deterministic knowledge and total ignorance) and its 'nature' (whether uncertainty primarily stems from knowledge imperfection or is a direct consequence of inherent variability). We have extended this typology - and the associated uncertainty matrix - by explicitly adding two additional dimensions (represented by columns) denoted 'qualification of knowledge base' and 'value-ladenness of choices'. These additional characteristics have also been brie y mentioned by Walker et al. 2003, as being specific features of knowledge-related uncertainty. Due to their importance for assessing and communicating uncertainties, we have decided to explicitly incorporate these dimensions in the uncertainty matrix as two additional columns. Moreover we have also slightly modified the location-axis of Walker et al. 2003, which was specifically designed for model-based decision support studies. Two novel location categories have been added, viz. 'expert judgment' and 'data', since these can often be clearly distinguished as separate identities apart from the other categories. Finally, the 'model-category' has been extended by classifying the original separate seperate categories 'inputs' and 'parameters' of Walker et al. 2003 as subcategories of the 'models'.