中文摘要：

在粗差定位一致的前提下，证明了拟准检定法与部分最小二乘法的粗差估值具有等价性；LEGE法粗差估值与部分最小二乘法粗差估值具有等价性，其条件为未受污染观测值等权独立，且与受污染观测值相互独立。仿真算例证明，当未受粗差污染观测值不等权时，拟准检定法与部分最小二乘法得到的粗差估值结果相同，均比LEGE法粗差估值结果更准确、精度更高。

英文摘要：

Detecting and manipulating multiple outliers is one of the most challenging topics in the research area of observation data quality control. There are four popular methods dealing with this issue, including data snooping, the method of simultaneous locating and evaluating multidimensional gross errors （LEGE）, quasi-accurate detection of gross errors （QUAD）, and part least squares（PLS）. The outlier estimation formulas of the four methods appear much different. It has been justified that the outlier estimations of data snooping and LEGE are equivalent under certain condition. Thus the efforts of this presentation are focused on discussing the outlier estimation equivalence of the last three methods. The preliminary assumption of the discussion is that the outliers have been successfully located. We have two conclusions. First, the outlier estimation of QUAD is equivalent to that of PLS. Second, the outlier estimation of LEGE equals that of PLS under the condition when the clean observation data （the first data group） has equal weights and is independent of the contaminated observation data （the second data group）. The simulated testing example reveals that if the clean data are not equally weighted, the outlier estimation of QUAD is equal to that of PLS, but both are more accurate and precise than the outlier estimation of LEGE.