Taxonometric evaluation

Transportation planning has for too long been unbalanced in that an excessive amount of time has been spent on building increasingly complex quantitative models at the expense of rather important aspects of the design and forecasting stages of the process. This paper therefore sets out to consider two related problems: that of forming a coherent and useful link between the design and evaluation stages and that of suggesting a method of dealing with probabilistic forecasts within the evaluation model.The solutions to both of these problems are linked. It is suggested that the structure of and variables contained within the formal evaluation model must also necessarily play a part in the scheme design or selection process. This process is itself seen as one of hierarchical selection wherein classes of scheme or solution are successively more explicitly identified as sensibly different and, where possible, the preferred class is identified. Central to this taxonometric procedure is the idea of sensible differentiation between classes whose characteristic performance is only known uncertainly and some algorithms for achieving this are suggested.