Re: st: Dropping observations in relation to mean and standarddeviation

[Friederike Ballaire <fhornber@mail.uni-mannheim.de>]

Hi Daniel,
thanks for your reply and sorry for not mentioning the cross-posting.

As I am doing a replication or rather an adaption of a previous study
I try to follow their steps as precisely as possible, and that's the
way they treat outliers in their study. But this might be an
interesting point for the discussion part.

I have already been thinking about reshaping my data but for the
actual data analysis I will need the data in the present form. I
didn't think about reshaping it, droping the outliers and reshaping it
again. Maybe that's an option? I see if that'll work.

Thanks and have a great day!
Friederike

On Tue, Jan 21, 2014 at 5:34 PM, daniel klein <klein.daniel.81@gmail.com> wrote:
> This a cross-posting
>
> http://www.stata-forum.de/stata-syntax-f18/drop-bei-bestimmten-wert-abhanging-vom-mittelwert-t662.html
>
> Please see the FAQ concerned with cross-postings:
> http://www.stata.com/support/faqs/resources/statalist-faq/#crossposting
>
> and Nick Coxs' discussion of the issue on Statalist:
> http://www.stata.com/statalist/archive/2013-06/msg00817.html
>
> Concerning your question, there have been lots of interesting
> discussions on the list about so-called 'outliers' and it seems the
> general advice is to think twice, whether these are really coding
> errors or the like, before deleting them automatically based on some
> rule of thumb (e.g. larger than two or three SDs).
>
> I have no exact idea about the structure of your data, but you might
> want to -rehsape- rather than -stack-? Probably -egen-'s -sd- function
> combined with a -by- prefix is a way to go from there.
>
> Best
> Daniel
>
> --
> Hi,
> I tested 144 people and now I need to drop observation that are more
> than three standarddeviations away from the mean of two variables for
> one subject.
>
> [...]
> Is there a special command for that?
> Otherwise I would just calculate the mean and standarddeviation for
> every subject and then calculate three standard deviations away from
> the mean and use the drop command.
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
>
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/