Absolutely. You get what you asked for,
a new temporary variable.
The bit that was not obvious to me when
I learnt this is that
tempvar t
assigns a new name each time. This will make it
difficult to use any previously assigned
names unless you keep track of them. So
a common idiom is
in some loop {
tempvar t
...
local tlist "`tlist' `t'"
}
That way, each time round the loop, the
latest name `t' is put in a bag with
all the previous ones. Thereafter,
you work with the `tlist'.
Nick
[email protected]
Cameron Hooper
> Michael Blasnik wrote:
> > I think you may need to add:
> >
> > drop `res'
> >
> > at the end of your loop.
>
> Hi Michael
>
> Thanks for your suggestion. I did not realize -tempvar-
> operated in this
> fashion. Added the -drop- command fixed that particular
> problem. (Others
> have since arisen!) Just to convince myself I wrote the
> following test
> program to explore the behaviour of -tempvar-. Other complete newbies
> might also find it interesting:
>
> . clear
> . set obs 5
> obs was 0, now 5
>
> . gen x = _n * uniform()
>
> . capture drop nv
> . capture program drop tmpv
>
> . program tmpv
> 1. qui {
> 2. gen nv = .
> 3. forvalues i = 1/5 {
> 4. tempvar t
> 5. gen `t' = x[_n]
> 6. describe
> 7. replace nv = r(k) in `i'
> 8. }
> 9. }
> 10. end
> .
> . tmpv
> . list
>
> +---------------+
> | x nv |
> |---------------|
> 1. | .5415702 3 |
> 2. | .0175157 4 |
> 3. | .921708 5 |
> 4. | 1.988945 6 |
> 5. | 4.485629 7 |
> +---------------+
>
> Yep. Each each iteration of the -forvalues- loop generates a
> new variable.
*
* 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/
