A SIMULATION METHOD FOR DETERMINING THE CONFIDENCE INTERVAL OF A WEIGHTED GROUP AVERAGE VALUE OF TIME

Over the past decades, value-of-time (VOT) measures have frequently been used for the appraisal of investments in road infrastructure and public transport facilities. It was only in the 1980's, however, that attention was paid to the confidence intervals of VOT measures. If the VOT itself is a construct of a single time and a single cost parameter, it is possible to determine the VOT confidence interval in a straightforward way for the estimation sample. This paper discusses a method of establishing confidence intervals of expanded VOT measures, in which both the time and the cost parameter interact with various segmentation variables. The expansion process, proposed but not completed for the early British study, involved the use of 'exposure measures' based on national statistics to expand the measured VOTs for the large number of market segments and contexts identified by the models. The procedure first involves the use of a Jack-knife estimation procedure to derive correct confidence intervals of model parameters in the context of repeated measurements in the SP survey, extending the early work. Secondly, a simulation technique is used, which simultaneously draws parameters from a multivariate normal distribution defined by the variance-covariance matrix of the estimated parameters. For the covering abstract, see IRRD 899083.