Request to Nick Cox please

[Euslaner]

Sorry to trouble you but I can't locate the short command that drops all string variables from a dataset. If you could repeat it I would be grateful.

Ric Uslaner

[Hemanshu Kumar]

You can do that with:

ds, has(type string)
drop `r(varlist)'

[Nick Cox]

Euslaner (Eric/Ric Uslaner)

It's really best to assume that you're addressing the forum and that if you have a simple straightforward question that many experienced people will know the answer (as exemplified excellently by Hemanshu Kumar 's answer) and also that many beginners might share an interest in the answer, so that the thread title here is at best a distraction.

The question otherwise raises another. Often one has many string variables but that usually means that those variables are just better ignored for some purpose rather than dropped altogether. Also, many string variables may mean that essentially numeric data have been read in as string variables, usually because as read from another place the data contain metadata and/or characters that can't be interpreted as numeric in Stata terms (e.g. NA as codes for missing). The usual remedy there starts with destring -- or sometimes a realisation that the data should be read in again but more carefully.