| Abstract: | The issue considered is whether the biases that result from assuming a single value of waiting time in variable-demand headway optimizations (and the demand models on which they are based) can have a significant impact on the results of the optimization. Marginal benefit functions based on binary probit demand models, and incorporating the alternative assumptions of either a single value of time, a discrete distribution of time values, or a continuous distribution of time values, are derived and compared for several cases. These comparisons indicate the possibility of significant errors in several cases, the most realistic of which is that of a highly heterogeneous market composed of a large segment of “choice riders” and a smaller segment of “transit captives”. |
