家戶對住宅的選擇可以視為一連串的決策過程，包括租買選擇、住宅型式、區位 等選擇決策。本研究依巢式多項 Logit 模型架構建立住宅選擇模型， 並透過對包容值 (inclusive value) 係數之檢定予推論住宅選擇之決策程序。 本研究詳細說明了巢式多項 Logit 模型之理論基礎及模型架構，並以 1980 年美國之公用個體抽樣資料 (the Public Use Microdata Sample, PUMS) 為基礎， 選擇聖地牙哥 (San Diego, CA) 都會區為 研究區進行實證分析。研究結果發現，家戶將租買選擇列為最優先的決策，其次決定住宅類 型，包括住宅型式 (透天或集合住宅 )、室內空間 (如房間數 )、住宅區位 (如市區或郊區 ) 以及住宅品質 (如建築物年齡、衛浴及廚房設備、空氣調節等 )，最後在各住宅類型中挑 選最適當之住宅單元。

　　　　　The process of housing choice could be a bundleof decision making including tenure choice, housing type choice, and location choice etc. This paper try to find empirical evidences to explain the decision process:the simultaneous decision or the sequential decision. the model that applied to this study is the well-developed logit model in discrete choice theory. By testing the coefficient of inclusive value in the nested multinomial logit model (NMNL), we are able to answer what the decision structure of housing choice would be. The data we used in this study is thep Public Used Microdata Sample (the PUMS) of San Diego metropolitan area from 1980 Census data. The final result is that the first prior decision made by household is the tenure choice and then followed by the "housing type' choice which is classified by four indexes:the housing stucture type (single unit or multiple unit), the number ofrooms, the location(city or suburban), and the housing quality level. The last decision would be to select the housing unit from the given 'housing type'.