二维升降法

 提出了一种二维升降法，该方法不仅适用于极小样本情况，而且还可以综合利用以往积累的试验数据和当前的试验数据来预测敏感性变量的百分位值（安全疲劳极限、安全冲击强度等）和百分率（可靠度、破坏率等）。与传统的只能利用当前试验数据的升降法相比，其可利用的信息量有了大幅度增加，所以，在精度相同的情况下，可以节省大量试件；而在试件数一定的条件下，又能提高预测精度。还给出了一个对比实例。

TWO-DIMENSIONAL P-AND-DOWN ETHOD

Up and down method was originally devised for sensitivity testing by W.J. Dixon and A.M. Mood and developed by some other experts, but all these traditional up and down methods can not use previous data effectively, and, besides, the estimates for variance are not exact. This paper establishes a two dimensional up and down method which is applicable for small samples and can be used for estimating percentiles and percentages of sensitivity variables by combining previous experimential data and current test data, and furthermore, gives the confidence limits of the percentiles and percentages. The information that can be used in the two dimensional up and down method is much more than that in the traditional methods, because this method can make full use of the current test data and the previous data, while the traditional methods can only use current test data. Thus, fewer specimens are required in the two dimensional up and down method and the result is more accurate for the same number of specimens. A contrast example is also given in the paper.