Without more information, better advice is difficult. But
suppose that a state code was always two letters that
appeared just before a numeric substring.
Where the first numeric character appears can be put
in a variable like this:
gen first_numeric = 0
gen length = length(oldvar)
su length, meanonly
qui forval i = 1/`r(max)' {
replace first_numeric = `i' ///
if first_numeric == 0 & inrange(real(substr(oldvar,`i',1)),0,9)
}
Then the state code can be extracted as
gen state = substr(oldvar, first_numeric - 2, 2)
Nick
[email protected]
Michael S. Hanson
> Actually, given the proposed solution it appears that
> the worry is not
> specifically "about 8", but whether the "State" code (a)
> always begins
> with the 4th character, and (b) always is just 2 characters long.
> Depending on the universe of potential values of "oldvar", there may
> exist applications of -assert- that can check these desired
> conditions.
> However, notice that the currently proposed application of -substr-
> should work as intended on values such as "CITNA34", "CITNA034", and
> "CITNA0340", but will fail on "CINA134" and the like.
Nick Cox
> > Nevertheless, your description "about 8" is worrying.
> > In any observations that have some other number
> > of characters, this solution may be incorrect. Try
> >
> > . assert length(oldvar) == 8
[email protected]
> >
> >> That worked beautifully, thank you. Just saved me five hours
> >> of banging
> >> my head against the wall.
> >
> >>> Try
> >>>
> >>> gen str State = substr(oldvar,4,2)
> >>>
> >>> This assumes that your original variable was string.
> >
> > [email protected]
> >
> >>> I have a variable that has about 8 characters per
> observation and I
> >>> need to take 2 characters (4th and 5th characters) and
> generate them
> >>> as new variables representing state codes. So for example, the
> >>> variable has as one of its value CITNA134. The
> characters I need to
> >>> extract are NA, which will be recoded as State code. How
> would I go
> >>> about doing this?
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
