| 摘要: | 存活分析理論中之Cox等比率危險模式以及分層Cox等比率危險模式應用於機車騎士紅燈怠速熄火意願之研究，透過條件評估法進行問卷設計調查，經過資料整理、模式構建、參數校估及模式配適檢定後發現，｢節能減碳認知｣、｢機車使用特性｣、｢健康認知｣以及｢個人屬性｣四大類變數確實可以有效解釋機車騎士紅燈怠速熄火意願。 回收共603份有效問卷樣本資料以SPSS軟體進行Cox模式配適，但以log-log存活曲線檢定後發現必須重新以Cox分層模型進行配適，才能將不符合等比率的風險因子納入其中。最後依照每週騎車天數分為兩群(兩模式)，各自建構分層Cox等比率危險模式，經過篩選檢定後，發現兩模式中各有5項與4項變數，其中｢熄火再發動汙染不增認知｣、｢熄火意願(油價上升)｣、｢依從熄火意願｣為兩模式皆有之解釋變數。 最後，本研究參考了金融學界使用Cox模式作為預警決策模式的方法，透過存活函數的推算以及適當的判斷臨界值，可以觀察出解釋變數的改變帶來的實際效益─熄火時間的增加，以模式一中的變數，熄火依從信念為例，計算更動其變數值後的熄火時間變化量，發現能使若不同意跟從他人熄火者提升到意見普通，共可以增加280秒的熄火時間，模式中平均每人增加0.98秒；若是從不同意改變觀點為同意，共可以增加400秒的熄火時間，模式中平均每人增加1.4秒。 Within The theory of survival analysis, Cox proportional hazard model and the stratified Cox proportional hazard model were applied in to investigate motorcyclists’ willingness to do idling stop on the red light by means of the contingent valuation method questionnaire survey. Through data compilation, model construction, parameter calibration and goodness of fit test for the model, the result shows that motorcyclists’ behavior of idling stop on red is significantly influenced by four major categories: "the awareness of energy saving and carbon reduction awareness," "the characteristics of using motorcycle," "the health Awareness," and "personal property." Using the data of a total of 603 valid questionnaires with SPSS software to construct the Cox models. Finally, according to the days of riding motorcycle per week, divide into two groups (two models) and construct each group to become the stratified Cox proportional hazard model. After the model testing and variable selection, there are five variables in model I and four variables in model II, moreover, there are three same variables, "turn off and then launch a pollution not by perception, " "the willingness to do idling stop (because of the oil price rising)," "the belief of following other motorcyclists to do idling stop," in the two models. Finally, consulting the study of the financial academic decision making using the Cox model as a model of early warning methods, through the survival function of the projections and determine the appropriate critical value, we can observe the change of variables to count the actual benefits ─ the increase of idling stop time. For example in model I, we can calculate the increase of idling stop time with the variable of “the belief of following other motorcyclists to do idling stop.” We find that the idling stop and the value of the variable shows a positive trend, that the total idling stop time increases 280 seconds, even 400 seconds (the average idling stop time per person increases 0.98 seconds, even 1.4 seconds) if the belief of following other motorcyclists to do idling stop could be escalated. |
