在這篇論文中，我們考慮了非完整二序列二元交叉實驗設計 (incomplete two sequence dual crossover design) 的統計推論。我們推導出人體內差異(within subject variabilities) 在相等及不等的條件下，處理 (treatment) 的有效估計值及其t統計量。這個研究結果顯示在人體內差異相等的情況時，我們可獲得精確 (exact)的有限樣本結果。反之，在人體內差異不等時，精確有限樣本結果則無法求之。然而，在溫和的條件 (mild condition)下，我們建立一些有限樣本的精確結果及大樣本結果。 Statistical inference on the effect of two treatments, based on incomplete data from a two-sequence dual crossover design, is considered. Efficient estimates of the treatment effect and the corresponding t- type statistics are derived for both cases of equal and unequal within-subject variabilities. In the case of equal within-subject variability, exact results for finite samples are obtained. When within-subject variabilities are unequal, there may exist no exact finite sample results. Under mild conditions, some exact finite sample results and some large sample results are established.
