Re: st: problem with forvalues loop

[Nick Cox <njcoxstata@gmail.com>]

Apart from advising -reshape long-, I recommend considering something like this

forval s = 1/15
        local S : di %02.0f `s'
        gen retgrade`S'13 = .
        gen resc`S'13 = .

        forval y=1/10  {
             replace retgrade`S'13=retgrade`y' if resubj`y'==`s' & rect`S'13==1
             replace resc`S'13=resc`y' if resubj`y'==`s' & rect`S'13==1
        .}
}


Nick
njcoxstata@gmail.com


On 25 February 2014 16:37, Holly Heard <hheard@rice.edu> wrote:
> Dear Statalisters (apologies for long post):
>
> I have data on exams that high school students take in 2013. Each student
> can take more than one exam. The data were originally long (one observation
> per exam), but I reshaped it wide, so that I have one observation per
> student. Variables pertaining to each exam are numbered 1 to 10, based on
> the maximum number of exams taken by a student (most students took 3-4
> exams, but one student took 10). There are multiple variables based on each
> exam number: retgrade1-retgrade10 (grade student was in when took exam),
> resc1-resc10 (scale score of exam), etc. I need to reorganize and rename
> variables based on the subject of the exam (algebra, reading, etc.), which
> is denoted by resubj1-resubj10, which is the subject category of the exam
> (1=Algebra I, 2=Algebra 2, 4=Chemistry). There are 15 exam categories,
> although as noted, a maximum of 10 exams were taken by any one student.
>
> I created a forvalues loop to go through each variable denoted by exam
> number, and generate new variables based on subject number. My code is:
>
> forvalues y=1/10  {
>         forvalues s=1/15  {
>         if `s'<10 {
>          local s 0`s'
>         }
>         gen retgrade`s'13=retgrade`y' if resubj`y'==`s' & rect`s'13==1
>         gen resc`s'13=resc`y' if resubj`y'==`s' & rect`s'13==1
>         .
>
>         }
> }
>
> Here, y denotes exam number 1-10, and s denotes subject number 1-15. I add
> 13 to the end of the new variables to denote the year of data collection,
> and to distinguish new vars from old vars. It is possible for a student to
> have taken an exam for a particular subject twice (if they didn't pass the
> first time), so I create new variables only if the student took the subject
> exam only once (rect`s'13==1).
>
> This loop runs many times, based on the many *(23851 missing values
> generated)* statements I get, but then I get the following error:
>
> retgrade0113 already defined
> r(110);
>
> So apparently the loop repeats itself at some point. I realize that y and s
> are not fully nested (that would be too easy), but I'm not sure how else to
> write the code to do what I want to do. I tried reversing the order of the
> forvalues commands (y within s), but get the same error. As far as I can
> tell, I should only have one retgrade0113 as long as rect0113==1. Can anyone
> tell me what I'm doing wrong and how to fix it? Thanks, Holly
>
> Holly E. Heard, Ph.D.
> Senior Social Scientist
> HERC Data Specialist
> Houston Education Research Consortium, MS-28
> Rice University
> 6100 Main Street
> Houston, TX 77005
> Phone: 713-348-3415
> Fax: 713-348-5296
>
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/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/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/