12. Application of Stated Preference Technique for Travel Choice Modeling
Date of Start

January 2006
School of Planning and Architecture, New Delhi (R)


Scope and Objectives

(i) To appreciate the state of the practice for mode choice modeling and to identify issues relating to choice modeling for a new system,

(ii) To study mode choice attributes of user through SP surveys.

(iii) To develop travel choice model for the case study corridors.

(iv) To validate the travel choice model using RP survey data and carry out sensitivity analysis.

The steps for this study can be grouped into eight broad stages. Extensive literature review on SP/RP methods, mode choice behavior models and SP experiments was carried out. Followed by data collection, which included primary surveys, which was done in two parts, in the first stage a pilot survey to establish the significance, variation in user response, and selection of attributes was carried out, and followed by the main survey. Partial factorial design was carried out. Secondary data was also collected about metro network, metro rider-ship details, existing development and proposals. The data collected was analyzed to establish metro corridor characteristics and user characteristics. Mode choice models were developed for Dwarka sub-city extension corridor and Delhi Noida corridor. This was followed by modeling choice probabilities for MRTS using binary logistic regression. Model validation and sensitivity analysis was carried out.
Findings and Conclusions
(i) Travel cost and in-vehicle time are most important attributes emerging from stated preference experimental design results.

(ii) Multinomial and Binary Logit models are the common state of practice for mode choice modeling in India and abroad, particularly in case of new transport mode.

(iii) Out-of-vehicle cost is more critical in explaining the preference for MRTS while users are willing to accept even more higher in vehicle cost (fare) for MRTS.

(iv) In-vehicle time and out of vehicle time are more important components for users selecting MRTS indicating higher emphasis placed by users on time saving than money saving.

(v) Stratification of the binary logit model by income group gives better model prediction than the single binary logit model for all the users.