Selecting a subset of ordered variables without typing them manually

[Marc Asensio]

Dear all,

I am trying to execute a fairly simple operation:

egen duration_desk=rowtotal(t_*_pagesubmit)

The issue I have is that with the code I just showed, I am selecting all my variables that include 't_*_pagesubmit' in their name, and I am only interested in a few of them that luckly are ordered. However, in-between there are other non-related variables that do not follow the 't_*_pagesubmit' formula.

Say I am interested in variables 't_1_pagesubmit' 't_2_pagesubmit' and so on until 't_10_pagesubmit'. I could use a code that selects from 1 to 10 but how can I do that without selecting the variables in-between them?

Of course I could simply type all the variables manually but there must be a simpler way around.
I do apologize if this has been answered, I honestly did not know how to search for such a problem.
Thank you for your time.

[Joro Kolev]

Have a look at
[U] 11.4 varname and varlists.

[Hemanshu Kumar]

I can think of two sorts of things that can help with knotty problems in identifying the right variable list:

1. if the variables to be included (or excluded) can be identified using some characteristics, you might want to use the -ds- command and pick up the local it stores in r(varlist). See

   help ds

2. You can also use the flexibility of macro lists (perhaps in combination with the above suggestion) to do set operations on these lists. See

   help macro_lists

   .

[Nick Cox]

If the * in

 t_*_pagesubmit 

is catching more than 1 2 3 4 5 6 7 8 9 10 then ideally we need to know what that more is to suggest a better approach. But

forval j = 1/10 {      
     local vars `vars' t_`j'_pagesubmit
}    

egen duration_desk=rowtotal(`vars')

is one more general solution.

[Marc Asensio]

If the * in

t_*_pagesubmit 

is catching more than 1 2 3 4 5 6 7 8 9 10 then ideally we need to know what that more is to suggest a better approach. But

forval j = 1/10 { local vars `vars' t_`j'_pagesubmit

}
egen duration_desk=rowtotal(`vars')

is one more general solution.

Thanks, that sounds like a solution but indeed there is more. I just simplified it for the example but the * is basically the name of the variables so there is no common trend. One can be 'f130200' and another 'deviceuse_h'.

[Nick Cox]

You have lost me now. There is no way that a wildcard like

 t_*_pagesubmit

will catch either of f130200 or deviceuse_h. Other way round, if #1 doesn't epitomize your real problem, what does?

[Marc Asensio]

You have lost me now. There is no way that a wildcard like

t_*_pagesubmit

will catch either of f130200 or deviceuse_h. Other way round, if #1 doesn't epitomize your real problem, what does?

Isn't there a way to say select t_*_pagesubmit from t_f130200_pagesubmit to t_deviceuse_h_pagesubmit for example?

If not then I guess I'll have to do a macro and that's it.

[Nick Cox]

I see what you want now. it's the intersection of two sets,

unab A : t_*_pagesubmit
unab B : t_f130200_pagesubmit-t_device_h_pagesubmit
local wanted : list A & B

A, B and wanted are here all local macros.

It is possible that

t_?_pagesubmit t_10_pagesubmit 

is what you want but in principle the ? could match many single characters not 1(1)9.