概述:在生物医学和社会心理学研究中采用样本几何均值估计、比较人口几何均值的方法十分普遍。然而,由于测量工具的检测局限,有时无法观察到测量的实际值。处理这个问题的一种常见做法是用较小的正值常数来替代缺失值,然后在这些填补数据基础上进行统计推断。然而,这种简单的填补方法对推论的影响还没有研究过。我们在本文中阐明了这种简单的填补方法可能会大幅度地改变一项研究所报告的结果,因此即使检测限非常小,也会使结果难以解释。 Summary: The use of geometric mean estimates of samples in biomedical and psychosocial studies and the comparison of population mean geometries are common. However, the actual value of the measurement can sometimes not be observed due to the limited detection of the measuring tool. A common practice to deal with this problem is to replace missing values ​​with smaller positive constants and then make statistical inferences based on these fill-in data. However, the impact of this simple method of padding on inferences has not been studied. We illustrate in this article that this simple method of padding can drastically change the results reported in one study, so that even with very small detection limits, the results can be difficult to interpret.