RE: st: RE: RE: left right string

["Nick Cox" <n.j.cox@durham.ac.uk>]

I commended your answers, so why look for critique? I would have recommended what you did given the initial question, namely to use -split-. 

But Tyler made clear eventually that he wants to go from numeric variable to numeric variables. That being so, string manipulations are not needed. End of my story. The real question is "to go from a existing numeric variable with value labels to new numeric variables based on those labels", as said. 

"There's another way to do it" is always fair comment. 

Nick 
n.j.cox@durham.ac.uk 

Martin Weiss

" Martin gave good answers to the original question, but from this end of
the thread it is clear that the real question is different."

I do not understand this critique. I gave answers to the original question
_and_ the modification requested in
http://www.stata.com/statalist/archive/2010-05/msg01039.html. If anything,
my solution is automatic while Nick`s requires examination of the -label-s.
Am I missing anything here? What is the "real question", btw?

Nick Cox

Martin gave good answers to the original question, but from this end of the
thread it is clear that the real question is different. 

Tyler wants to go from a existing numeric variable with value labels to new
numeric variables based on those labels. 

If that is it, then you could just look at the labels and go (e.g.) 

gen age_min = cond(hhh_age == 1, 15, 
              cond(hhh_age == 2, 20, 
		  cond(hhh_age == 3, 25)))

OR 

gen age_min = 15 * (hhh_age == 1) + 20 * (hhh_age == 2) + 25 * (hhh_age ==
3) 

OR 

gen age_min = 15 if hh_age == 1
replace age_min = 20 if hh_age == 2 
replace age_min = 25 if hh_age == 3 

This is not an exhaustive list, no doubt. My point is just that you can look
at the list of values and value labels and work out -generate- and -replace-
code. 

Tyler's real example is no doubt a bit more complicated but the principle
should apply. 

Nick 
n.j.cox@durham.ac.uk 

Martin Weiss

" I just realized my ranges are labels"

No big deal, just -decode- the thing beforehand:


***********
clear*

set obs 3
gen byte hhh_age=_n

la def ages 1 "15-19" 2 "20-24" 3 "25-39" 
la val hhh_age ages

decode hhh_age, gen(newhhh_age)

split newhhh_age, gen(age) parse(-) destring
rename age1 age_min
rename age2 age_max

list, noo
***********

Tyler Frazier

I just realized my ranges are labels

On Tue, May 18, 2010 at 11:08 PM, Martin Weiss <martin.weiss1@gmx.de> wrote:
>
> <>
>
>
> You can generate the new variables and -destring- (which I assume you also
> want) in one fell swoop:
>
>
> ***********
> clear*
>
> inp str5 hhh_age
> 15-19
> 20-24
> 25-39
> end
>
> split hhh_age, gen(age) parse(-) destring
> rename age1 age_min
> rename age2 age_max
>
> list, noo
> ***********

Martin Weiss

> Have you had a look at -split-?

Tyler Frazier

> simple question, I hope -- I have a variable which is a range of ages,
> for example  the variable name is hhh_age and the quantities are
> 15-19, 20-24, 25-39 etc...
>
> and I want to generate a new variable (age_min) that returns the left
> two characters from hhh_age , and another variable that generates the
> right two -- (age_max)
>
> I searched but couldnt find it, seems like a very simple thing, but
> cant figure it out.
>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/