Abstract

We note that log transformations can be problematic when the variance of the underlying distribution is other than very small. We illustrate these problems in terms of lognormal sampling issues, interval estimation of the mean, and comparison of lognormal and logbinomial distributions with similar means and variances.

中文翻译：

---

日志转换：当你期待什么时不要期待

摘要

我们注意到，当基础分布的方差非常小时，对数转换可能会出现问题。我们通过对数正态抽样问题、均值的区间估计以及具有相似均值和方差的对数正态分布和对数二项分布的比较来说明这些问题。