Interval estimation of location difference with incomplete data
提出了一种渐近分布自由的置信区间，用于估计两个相关变量在数据不完全时的位置差异，该区间形式简单且比已知参数方法更稳健。
An asymptotically distribution-free confidence interval for the location difference of two correlated variables with incomplete data on both responses is proposed. The confidence limit has an explicit form and can be easily obtained. The new procedure is also shown to be more robust than other known parametric interval estimators.