Dropping many variables if all observations contain a certain text

[Lola Brahimi]

Dear All,
I am having trouble with a command in Stata and no matter how much I researched beforehand, I could not find a solution. I have a dataset that contains cross sectional data. There are around 1500 variables, where many of them contain from observation 1 to observation 29000 the value -1, i.e. ''missing''. I am trying to create a command that tells Stata that if a variable has only the value -1 from the first observation until the last one, then it should drop this variable and this should be done for all the variables that have this problem. Does anyone know how to proceed?

I thank you all for your time and your opinion.
Best.
Lola

[daniel klein]

This is not tested

foreach var of varlist * {
    summarize `var' , meanonly
    if (r(min) != -1) continue
    if (r(max) == -1) drop `var'
}

Perhaps even quicker

foreach var of varlist * {
    capture assert `var' == -1 , fast
    if (!_rc) drop `var'
}

Best
Daniel

[Lola Brahimi]

Dear Daniel,
It worked! Thank you a lot, you saved my day!

All the best,
Lola

[Nick Cox]

Daniel Klein's solution is excellent. Also with findname (Stata Jourmal)

findname, all(@ == -1)

would find all such variables after which

drop `r(varlist)'

would clear them out of the way.

[Lola Brahimi]

Dear Nick,
Thank you for the suggestion. I will try it as well.

All the best,
Lola