Calculating trimmed mean

[Dylan Gadhia]

I am trying to calculate trimmed mean values, where the largest and smallest 5% of returns are eliminated, in order to remove outliers from my dataset. Can anyone suggest how to go about doing this? Stata doesn't recognise the trimmean command. Thanks, Dylan

[Sergiy Radyakin]

Stata doesn't recognise the trimmean command

😊😊😊

That's because Stata is not Excel and uses different language for commands.
Btw, technically treammean is not a command, but a function.

sysuse nlsw88, clear

centile hours , c(5 95)
summarize hours if inrange(hours, `r(c_1)', `r(c_2)')

summarize hours // for all - for comparison

Highlighted lines calculate the mean removing 10% of the data (5% of the smallest and 5% of the largest).
If you don't have any ties, missings, etc this is probably all that is required.

[Nick Cox]

Removing outliers is not what a trimmed mean does. It just ignores values as specified in the tail(s) of a distribution. Any ignored values remain in the dataset.

trimmean (Stata Journal) is a community-contributed command. As such, you must install it before you can use it.

search trimmean, sj

should open a viewer window and clicking on st0313 will start installation. Or you can install it from SSC, using

ssc install trimmean

At the time of writing, it's the same code and essentially the same help file. at the Stata Journal and SSC sites.

EDIT I didn't see Sergiy Radyakin's post before. But I can assure everyone that trimmean is a command, as documented above.

I don't know of any treammean() function.

[Sven-Kristjan Bormann]

I think that Sergiy Radyakin was refering to an Excel function with the name treammean() and not a Stata function.

[Nick Cox]

Sven-Kristjan Bormann That comment led me to Google and find that Excel does have a TRIMMEAN() function. It's still a guess whether Dylan Gadhia was trying to apply similar syntax in Stata as #1 didn't show any syntax.

This is, as you will appreciate, precisely why we have lots of detailed advice in the FAQ, including showing the code you tried. https://www.statalist.org/forums/help People who think they're too busy to read and act on it often find that good answers arrive slowly and painfully, if at all.

[Sergiy Radyakin]

I think that Sergiy Radyakin was refering to an Excel function with the name treammean() and not a Stata function.

Exactly. Since the topic starter didn't clarify what treammean() (s)he wanted, and the Excel function popped out as the most prominent result.

The code that I posted matches the example from the Excel documentation:

clear
input x
4
5
6
7
2
3
4
5
1
2
3
end
centile x, c(10 90)
summarize x if inrange(x, `r(c_1)', `r(c_2)')

    Variable |        Obs        Mean    Std. Dev.       Min        Max
-------------+---------------------------------------------------------
           x |          9    3.777778    1.394433          2          6

With radio silence from the topic starter it's anybody's guess what was meant indeed.

Best, Sergiy

[Nick Cox]

Sergiy Radyakin I share your frustration, but again no treammean() is in sight any where but in your posts. I've taken a strong interest in trimmed means and their literature and I've never seen the term treammean used anywhere for anything, although perhaps something like that is in use in Russian or some related language.

Dylan Gadhia did "like" my post, but at most the person whose post is liked knows who did the liking. If a "like" is intended as thanks, then the thanks are private.

https://www.statalist.org/forums/help#closure

16. How should I maintain threads I start?

16.1 Close by giving a summary and thanks

Trying to wrap up a thread you started is helpful, especially if you report what solved your problem. You can then thank those who tried to help. Conversely, ignoring answers is less sociable, even if those answers did not solve your problem. "Thanks in advance" does not absolve you from either expectation.

Please note that a Like on a post is not publicly visible as coming from you and, while friendly, also does not absolve you from either expectation.

[Attaullah Shah]

The correct spelling of the function is trimmean() in Excel. https://support.microsoft.com/en-us/...a-63d988f511d3