[Nick Cox]

Sorry, but I don't understand enough about your data to offer further advice.

[Juan Miranda]

Nick, do not worry anymore. Thank you, you help me a lot, and this community too.
A hug.

[Ovidiu Rotariu]

Hello Nick,
I have the following problem

destring clinmeas_esr, gen (esr_new) force
clinmeas_esr already numeric; no generate

distring works as long as there are values which are strings in "clinmeas_esr" , however there are instances when all values are missing and distring treats "clinmeas_esr" as numeric.

Any solution around it?
many thanks,
Ovidiu

[Nick Cox]

Sorry, but I see no problem here. You want to change your dog into a cat, but then it turns out that it is already is a cat.

However, if it is really true that all observations in a variable are missing, then the variable is useless any way.

If I am not understanding correctly, I think you'll need to give more details.

[Ovidiu Rotariu]

I'm looking to variable "clinmeas_esr" in 25 .dta files (all files have the same name/format, as collected longitudinally). In most of them there are "accidentally" string data and this is why I use "destring clinmeas_esr, gen (esr_new) force" to force the non-numeric into missing. However, in 3 of the 25 files there is no data in "clinmeas_esr" and this when I get the "no generate" message

many thanks,
Ovidiu

[Nick Cox]

Ok, so not a problem. There was nothing (useful) to do.

[Ovidiu Rotariu]

Hi Nick,
I found a solution using: https://www.stata.com/statalist/arch.../msg00529.html
many thanks
Ovidiu