Data from 44 transport studies undertaken in this country between 1962 and 1971 were used to make the first comprehensive assessment of the predictive accuracy of British transport studies by comparing the study forecasts with the actual changes that have taken place. Scheme-specific studies such as inter-urban highway appraisals, rural by-pass studies, traffic management schemes and certain transport studies such as those for bus-rail interchanges, were excluded. It was found that nearly all of the forecast items considered were overestimated. On average, population was overestimated by 10 per cent, car ownership and household income by 20 per cent, and highway and public transport trips by 30 to 35 per cent. Traffic flows across highway screenlines were overestimated by an average of 13 per cent. An assumption of zero change in the input parameters would not have produced markedly greater forecast errors in any of the items and in many the average errors would have been considerably less. There was no evidence that the more recent transport studies or those that used more sophisticated modelling techniques (up to 1971) performed better than the others. Although the highway trip forecasts were dominated by the errors in the planning variables, there is evidence that the models used for forecasting public transport trips contained errors of specification. However, despite the general over-estimation of the main parameters there was some measure of agreement between the rankings of the observed and forecast changes. (Author/publisher)
Samenvatting