應用屬性基礎敘述性選擇法於汽車通勤者路線選擇行為之研究

本研究應用屬性基礎敘述性選擇法，並以Hensher研究之問卷調查為藍本，進行適合台灣特性之電腦問卷設計，問卷內容依照替選方案數量、選擇集合數量（情境數）、屬性數量、屬性水準數量及屬性水準範圍等多重維度變異進行實驗設計，藉此蒐集台中市小客車通勤者路線選擇行為之相關資料，進一步探討不同設計維度及屬性處理規則對通勤者路線選擇行為之影響。本研究建構設計維度交叉因子模式，並採用混合羅吉特建構屬性處理策略之模式，同時建立旅行時間節省之願付價格，最後並與澳洲研究結果相比較。由願付價格分析結果發現設計維度因子中屬性數量、屬性水準數量、情境數量及屬性值範圍等具有顯著差異，與澳洲結果不完全一致；而屬性處理策略方面，不考慮屬性處理策略，造成其願付價格估計顯著偏差，與澳洲結果一致。

關鍵字

敘述性選擇法, 混合羅吉特, 路線選擇, 旅行時間節省價值, 屬性處理規則, 設計維度

並列摘要

This study was based on attribute-based stated choice method. Experimental design, adopted from Hensher’s study, was varied with variation of design dimensions, such as numbers of alternatives, numbers of choice sets, numbers of attributes, numbers of attribute levels and range of attribute levels etc. Computer aided personal interview surveys can display multidimensional scenarios to participants to collect the data of commuters’ route choice behavior in Taichung city, Taiwan. Several multinomial logit models were applied to investigate the interaction effects between design dimensions and time attributes on commuters’ route choice behavior. Mixed logit models were also applied to investigate the attribute processing strategy of commuters. The willingness to pay calculated from the estimation results, both Taiwan and Sydney, can be compared to gain insightful results between this two countries. Results indicate that number of attributes, number of attribute levels, number of choice sets and attribute range have statistically significant influences. It also implies that willingness to pay without attribute processing strategy has statistically significant bias.