- 學位論文
實體書與電子書個體選擇模式之建立與分析
Establishment and Analysis of Individual choice model For Physical book and E-Book
摘要
電子書隨著科技普及、環保意識的高漲以及政府積極推動之下，電子書已成為日常生活中容易接觸到的另類書籍。國內對於電子書之個體選擇行為，過去僅高如瑩(民98)運用二項羅吉特模式加以探討；然該研究並未探討個體異質性對於受訪者之影響，因此本研究分別採用羅吉特與混合羅吉特模式來建立模式，並探討在不同的社會經濟背景與生活型態之下，兩種模式之解釋能力與預測能力是否有差異。 本研究採用高如瑩(民98)問卷資料，分別以羅吉特與混合羅吉特模式來建構含有方案屬性、社經特性、生活型態變數、以及購買意願變數之基礎模式。為了驗證羅吉特模式是否可以表現出足以媲美混合羅吉特模式的效果，因此進一步在效用函數中加入方案屬性與社會經濟變數等的交互作用；參數校估完成之後，再固定參數值代入99年收集的資料進行預測。 研究結果發現，混合羅吉特模式在資料配適度上優於羅吉特模式。然而在預測能力方面，若以概似函數為評估指標，羅吉特模式全面優於混合羅吉特模式；當以正確預測百分比來評判時，混合羅吉特模式才有稍好一些的表現。此結果說明，如經由詳細的市場區隔來建立模式，在實際預測上，羅吉特模式的表現並不比混合羅吉特模式差，此發現在實務應用上頗具有參考之價值。
並列摘要
With the popularization of IT technology, arising of environmetal conciousness, and the impetus of our government, E-BOOK has gained more attentions nowadays, and has become an special form of reading. In the past, there's only one research that discussed about Individual choice model of E-Book in Taiwan( Gao,2009). However, it didn’t make a discussion on heterogeneity for consumers. Therefore, this study uses Mixed Logit model and Logit model separately to establish models, and to discuss wether both the explanation and forecast ability of the two type models have differences under different economic variables and life styles This study adpots the data collected by Ru-Ying Gao(2009), using Logit and Mixed Logit to construct model which product attribute, economic variable, life style and purchase factors are added. For verifying the result of Logit model to see if it can work as well as Mixed Logit model, we add the interactions of product attribute and economic variable into the utility function. After all the parameters are calibrated, we set the parameters fixed and use the data collected in 2010 to make predictions. The results of the study indicates that the Mixed Logit model is better in fitting the data. As to forecasting ability, Logit model will be better than Mixed Logit in every facets if we set likelihood function as the evaluation indicators; When using the percetage of predicting correction as the assessed indicator , Mixed Logit may perform well in prediction. This result shows, in the prediction, if the model is established with detailed market segmentation, Logit model will have a nice performance and it will not be worse than Mixed Logit model. This finding is considered valuable in practical applications.