當雙截切(double-truncation)發生於在壽命資料分析時，我們蒐集到的個體失效時間若且唯若落於某個特定的時段內，而該截切的時段(也就是左截切和右截切的時間)因抽樣設計所影響。本篇論文研究壽命變數T於雙截切的情況下，探討對數-位置-尺度模型(log-location-scale model)及加速失敗時間模型(accelerated failure time model)。我們基於概似推論提出估計模型參數的方法，進而利用牛頓-拉弗森演算法(Newton-Raphson algorithm)得到點估計及信賴區間、信賴區域(confidence region)及信賴帶(confidence band)等區間估計。我們利用模擬實驗來查驗所提出方法的準確性，最後以現場可靠度研究(field reliability study) ─ Equipment-S data作為例證。;Double-truncation appears in the lifetime data analysis when the units are collected if and only if their failure occurs within a certain timespan. The timespan is defined by a left-truncation limit and right-truncation limit specified by the sampling design. This thesis studies the lifetime variable under the log-location-scale model and the accelerated failure time model when is subject to double-truncation. We develop likelihood-based inference methods for the parameters in the models. In particular, a Newton-Raphson algorithm is developed for point estimation. Confidence interval, region and band are developed for interval estimation. We conduct simulation studies to examine the accuracy of the proposed methods. The illustrations of the proposed methods are given by real data from a field reliability study, Equipment-S data.
