Developing a Successful Stated Preference Methodology for Determining Destination Choice Coefficients and Using It to Investigate Its Empirical Structural Relationship with Toll Route Choice



Developing a Successful Stated Preference Methodology for Determining Destination Choice Coefficients and Using It to Investigate Its Empirical Structural Relationship with Toll Route Choice

Authors

Peter Davidson, Peter Davidson Consultancy, Collins Teye-Ali, Peter Davidson Consultancy, Rob Culley, Peter Davidson Consultancy

Description

This paper traces the different approaches adopted in a series of stated and revealed preference exercises in Nigeria in an attempt to understand this structural relationship between destination and toll route choice and reports on the final successful approach.

Abstract

Extensive work exists in the literature and in practice about the structural relation between mode and destination choice models ie whether mode is above destination or destination above mode choice. In the US for example, destination choice is considered to be less sensitive to changes in transport level of service variables than mode choice whilst the reverse is largely true in the UK and this result is found even when these models are estimated simultaneously. However, very little work exists on the structural relationship between destination choice and toll route choice and the magnitude of the structural parameters. This is important because it could have a profound effect on the model's traffic forecasts and (in the toll road context) the revenue forecasts.

To investigate this, would either need revealed preference data from existing toll road situations or stated preference data. The literature on the analysis of such revealed preference data is sparse and inconclusive. On the other hand, it has hitherto not been possible to apply stated preference to destination choice for practical reasons such as for example that destination choice would need as many alternatives as destinations and stated preference is limited by its ability to deal with more than a few alternatives.

However during the course of several successive stated and revealed preference surveys, we have developed a successful experimental design and survey instrument so as to measure destination choice coefficients. We have taken this further so as to deal with the combined choice of destination and route choice where route choice is in the context of a toll route - so one choice is generally a high quality tolled route and the other is the best untolled main road alternative. We have used this stated preference and revealed preference experimental design to investigate whether destination choice is above route choice or vice versa.

This paper traces the different approaches adopted in a series of stated and revealed preference exercises in Nigeria in an attempt to understand this structural relationship between destination and toll route choice and reports on the final successful approach. It presents the detailed estimation results from the successful survey together with the estimated structural parameters using simultaneously estimated nested logit model estimation.

These results were put into our model of Nigerian toll roads and used to forecast the traffic and revenue on a toll road. These forecasts were compared with the forecasts from different types of model structure so as to illustrate the effect of proper nesting structures on the precision of toll road forecasts of traffic and revenue. We found that the relative positions of destination and route choices and the magnitude of the structural parameters had a significant impact on the estimated optimal toll value, forecast traffic and forecast revenues.

Publisher

Association for European Transport