不知道手册里的说明有无回答到您。但我觉得可以参考看看。

Example 6: Oversampling observations
bsample requires the expression in exp to evaluate to a number that is less than or equal to the
number of observations. To sample twice as many male and female patients as there are already in
memory, we must expand the data before using bsample.For example,

use http://www.stata-press.com/data/r11/bsample1, clear
set seed 1234
expand 2
bsample, strata(female)
tab female

如果手册的回答不能满足您的要求，或者我引错了，请见谅，
那就请牛人或神人补充好了！

谢谢啦，我先用用哈
但是有个问题哎，就是这种把样本扩大2倍之后有重复的抽取和在原样本（样本数为n）中有重复的抽取2n次相比，结果是不是会不一样啊？会差的很大吗？