real() function and destring command

[johnkim]

Dear all,

Hi, I have two questions regarding real() function and destring command.
I have string variables that are actually numbers, e.g., "230,000". If I am correct, real("230,000") generates a missing value and doing destring is not possible (unless you command ignore).

My first question is, can Stata recognize "230,000" is a number? Quintessentially, I want a function like real(), except that it will NOT generate a missing value and rather generate 230000.

Now, suppose I have multiple variables, assets196001, assets196001, ... assets201412 (i.e., assetsYYYYMM). I tried, destring assets*, gen(destr*), but it seems that gen(destr*) is the problem. I'm guessing that I cannot use * in gen() to refer to a pattern in the reference variable names.

My second question is, can I somehow use * trick to generate new multiple variables, or do I have to literally type gen (destr196001 destr196002 destr196003... destr201412)?

I'd greatly appreciate your advice and guidance. Thank you very much in advance!

Best,


John

[Maarten Buis]

Why do you not want to use the ignore()? That option is exactly designed for your purpose.

On specifying multiple new variables see: help newvarlist

[Daniel Bela]

I think Maarten is right, have a look at the ignore option of destring in the help. What you need should be:

destring varname , generate(newvarname) ignore(",")

This should, more or less, produce the same result as manually removing all commas and converting using real() afterwards:

generate newvarname=real(subinstr(varname,",","",.))

-destring-, however, can take care of multiple variables at once.

Regarding the list: you can easily construct a macro containing the names of all new variables you need, without manually typing them explicitly:

unab sourcevarlist : assets*
local targetvarlist : subinstr local sourcevarlist "assets" "destr" , all
destring `sourcevarlist' , generate(`targetvarlist')

Regards
Bela