Exploring area- and individual-level determinants of mode choice - A multilevel modeling approach

  • 林 羿汝

Student thesis: Master's Thesis

Abstract

Getting more understanding for individual mode choice in different areas is an appropriate way to support us to make transportation policies to reduce traffic congestion air pollution and provide a friendly transportation service As scooter mode has been received more attention we investigated the impact of individual- and area-level determinants on mode choice relative to reference mode (private vehicle or scooter) in Taiwan to provide some reference information for individual mode choice behavior Multinomial logit model has been used to analyze mode choice data However neglecting the possible existence of hierarchical structures with nested data could reflect limitations for the estimation Multilevel model can overcome these limitations then reduce statistical bias and improve statistical power Thus this research used multilevel models to identify the hierarchical-level determinants of mode choice in Taiwan The study sample included 24 832 respondents with living in 20 cities/counties and was collected by Taiwanese Ministry of Transportation and Communication (MOTC) in 2012 Data on individual-level were provided from the department of Statistic MOTC and city-level data were provided from the Statistics Department of Ministry of Interior National Development Council and Directorate General of Highways MOTC Results show variations for hierarchical-level determinants of mode choices relative to reference mode And it also verifies that the outcome variables expressed in multinomial logit model can also be used in multilevel model analysis Individual mode choice behaviors are associated with individual characteristics and area features where he/she lives Most variables at individual-level have significant influence on each mode choice relative to reference mode Among variables at area-level each mode choice will be influenced by different determinants relative to reference mode In the end we hope these finding may be useful for policy-maker to make transportation policies to change individual mode choice for reducing traffic congestion air pollution
Date of Award2014 Jul 29
Original languageEnglish
SupervisorChing-Fu Chen (Supervisor)

Cite this

'