應用階層線性模式於通勤旅程脈絡效果之研究

The Application of Hierarchical Linear Model to Exploring Contextual Effects within Commuting Journeys

廖祐君(Yu-Chun Liao) ； 楊士賢(Shih-Hsien Yang)

《先進工程學刊》 5卷4期 (2010/10) Pp. 357-365

https://doi.org/10.29948/JAE.201010.0010

摘要

基於消費者行為理論，旅運者為追求最大效用(utility)，除透過參與活動累加效用外，為減少旅運時所帶來之負效用，傾向以鏈結活動之方式參與，以減少返家／出門之重複途程，因而將不同活動/旅次與工作鏈結，結合成通勤旅程。當以鏈結旅程觀點思考旅運者之活動產生(activity generation)時，若以傳統之迴歸分析探討，將可能忽略自變數間之階層關係（旅次及旅程層級），而難以深究階層性之影響程度。因於旅運行為資料之層級特性，本研究採用階層線性模式（Hierarchical Linear Model, HLM，或稱Multi-level analysis），即用以解決自變數本身具層級架構之脈絡效果 (contextual effects)對依變數之影響問題；可檢視高階旅程層級與低階旅次層級變數對活動產出之影響，並透過不同層級變數之脈絡效果反映層級性質與所屬關係，藉此進一步解釋旅運者每日活動產生之非同步決策行為。研究結果顯示，階層線性模式對於旅運變數之階層序列性有良好之辨識效果外，亦能有效析離脈絡效果之影響。

關鍵字

階層線性模式 ； 脈絡效果 ； 旅運行為 ； 旅次鏈

並列摘要

Based on economic theory, commuters' activity-travel behavior tends to pursue maximum utility in their daily schedule. Therefore, commuters are willing to chain trips to reduce disutility via cutting repeating home-side traveling. Trip-chaining behavior of urban commuters has been modeled sequentially or simultaneously to establish various models of decision-making, including the activity generation model. It is well recognized that how complex the activity/trip generation model is, not to mention the trip-chained journey generation model. Instead of using ”trip” as analytical unit, the state-of-art studies then focus on modeling with ”journey” trying to capture the effects of trip-chaining related variables in addition to traditional disaggregate-based attributes. However, the contextual effects between journey-and trip-levels were ignored by previous studies and therefore never examined. To tackle the problem, this study used Hierarchical Linear Model to examine the contextual effects between variables during the complex activity-travel decision-making procedure. The results confirmed our hypothesis that trip- and journey-variables exhibit different leveled influences, and the contextual effects were significantly shown. Based on this study, the findings are helpful to improve our knowledge about commuters' activity-travel behavior and the related travel or land-use policies for planning.

並列關鍵字

Hierarchical Linear Model HLM ； contextual effect ； trip-chaining ； travel behavior