why the frequency of variables would not change after winsorized?

[Zoey Gu]

Unlike the trim, which remove the extreme data, the winsor function would replace the extreme data ( like the 1% smallest or the top 3 highest, bottom 3 lowest) with the data at required percentage.

[Nick Cox]

Trimming consists of ignoring specified values in one or both tails: nothing is removed in any Stata implementations I've looked at.

Sorry, but I don't understand what you're asking about WInsorizing. There are a winsor command and a winsor2 command on SSC, but winsor is not a function.

[Zoey Gu]

Trimming consists of ignoring specified values in one or both tails: nothing is removed in any Stata implementations I've looked at.

Sorry, but I don't understand what you're asking about WInsorizing. There are a winsor command and a winsor2 command on SSC, but winsor is not a function.

Hi Nick, thanks for reply! I used to think that winsor solve the outlier problem by removing the data from the sample observations, but I noticed the obs are not changed, so I didn't know how does winsor work. And then I've learned it replaced the outliers with the data at the required percentage. (Like the three smallest numbers are 0 1 1, and the second smallest numbers are 10, then the winsor will replace 0 ,1, 1 all with 10. ) It makes sense to me. Thank you!