多個層級、關聯隨機效果邏輯迴歸模型應用於探討休閒旅遊者習慣性的運具選擇行為及其影響因素

Multilevel Logistic Regression Modeling with Correlated Random Effects: Application to the Choice of Mode for the Leisure Traveler

應用多個層級、關聯隨機效果邏輯迴歸模型，分析休閒旅遊者習慣性的運具選擇行為及其影響因素，不僅可考量不變的組間共變項，也可考量隨時間改變的組內共變項。以臺灣北、中、南及東四區主要車站與休息站，選擇搭乘火車與自小客車的1,162位旅客為研究對象，蒐集人口基本統計變項及最近3次國內觀光旅遊選擇交通工具的情況及其10項主要考量因素。結果發現，除了婚姻、教育與職業等組間共變項之外，運具選擇行為仍會受到時間長短、付款方式、節能減碳、服務態度、班次多寡、舒適程度、安全性與機動性等組內共變項的影響。但是習慣性的運具選擇行為僅受到距離遠近、班次多寡及機動性的影響。這種方法與這項發現將可提供未來休閒旅遊者習慣性的運具選擇行為研究及其改變策略的參考與依據。

The purpose of this study is to apply multilevel logistic regression modeling with correlated random effects to explore the traveler's mode-choice behavior and its influential factors. This type of model can be used to assess the influence of between-subjects covariates and within-subjects covariates on the correlated dichotomous response variable. A total of 1,162 travelers, who took a train or drove a car, were invited to participate in this study at the major train stations and highway rest stations in Taiwan. They were asked to report their three most recent leisure journeys and their ten major considerations about the choice of mode. The results show that in addition to the traveler's marriage status, education, and occupation, his/her mode-choice behavior is influenced by his/her major considerations, such as travel time, the form of payment, energy conservation and carbon reduction, quality service, train runs, comfort, safety, and flexibility.Nevertheless, the traveler's habitual mode-choice behavior is influenced only by travel distance, train runs and flexibility. Therefore, these findings will provide us with the influential factors of the traveler's habitual mode-choice behavior, and this type of model will be applied further to explore these kinds of related issues.