Re: st: Use egen command with time series operator

[Nick Cox <njcoxstata@gmail.com>]

I add some further comments to this very good advice from Maarten.
First, one example given by Catharina was

egen mean_x1 = mean(l.x1)

Firing up -egen- to get a single constant, the mean, is a poor idea
compared with

su l.x1

which gives you that constant. If you want the mean to be in a variable then

su l.x1, meanonly
gen mean_x1 = r(mean)

is the way to go.

Very probably that example was just by way of illustrating the -egen-
problem, but I didn't want it to pass by without comment.

Second,  Catharina wrote

"Please consider that it is not reasonable for my programming to first
generate the lag/lead and then generate the mean."

I've got to ask: Why do you say that?

Nick
njcoxstata@gmail.com


On 28 November 2013 09:55, Maarten Buis <maartenlbuis@gmail.com> wrote:
> On Thu, Nov 28, 2013 at 9:03 AM, Klepsch, Catharina <klepsch@bwl.lmu.de> wrote:
>> I was wondering whether it is possible to use the egen command with time series operators.
>> For example, I tried to generate the mean value for a lagged variable and used:
>>
>> sort ID YEAR
>> egen mean_x1 = mean(l.x1)
>>
>> Unfortunately, there occurs an error message which states:
>> not sorted r(5)
>>
>> I'm not quite sure why this happens, because I sorted the data before. Even if I type:
>> bys ID YEAR: egen mean_x1 = mean(l.x1)
>>
>> the same error message occurs.
>>
>> Any ideas how I can fix that? Please consider that it is not reasonable for my programming to first generate the lag/lead and then generate the mean.
>
> Many of the -egen- commands sort the data internally, thus overwriting
> your sort prior to calling the -egen- command. This is also true for
> the -mean()- egen command, as you can see when typing -viewsource
> _gmean.ado-. This is why the help-file of -egen- contains the warning:
> "Explicit subscripting (using _N and _n), which is commonly used with
> generate, should not be used with egen". So you cannot do what you
> want with the -egen- command. However, if you look at the source code
> of the mean() egen-command (the -viewsource- command mentioned above),
> you can see tricks on how to compute this from first principle.
>
*
*   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/