Transformations and the Jackknife
本文从统计角度分析刀切法何时适用，将其视为带伪误差的定位参数模型的最小二乘估计，提出通过变换使误差方差恒定，从而为一般问题找到刀切前的变换。
SUMMARY Whether to jackknife a parameter estimate or some function of it has in the past relied on recommendations based on folklore rather than statistical analyses. This paper puts the jackknifed estimator into the context of finding a least squares estimator from a model involving a location parameter with additive “pseudo” errors. Obtaining and interpreting confidence intervals from the partitioning of the “total” sum of squares into “residual” sum of squares, plus a sum of squares due to a hypothesized value of the parameter, requires the error variance to be constant. This in turn can equally be interpreted as requiring the variance of the original estimator to be constant. Therefore, using this criterion of jackknifing “pivotal quantities”, it is possible in principle to find a prejackknifing transformation for any general problem.