Re: st: -foreach- loop to fit regression at each level of a variable

[Milena Przheska <milena.przheska@cosmicdevelopment.com>]

It appears my messages bounced, so I have to send this again.

Dear Sergiy, Klaus,

Thank you very much for your help.

Sergiy definitely hit a bull's eye! The code worked like a charm.

The reason why I did not want to use interactions was because I was
afraid it might interfere with the estimates of the other predictors
or cause a problem of multicollinerarity, while my only intention was
to check out the robustness of the model.

So bottom line, I used Sergiy's code along with the command -eststo-
from the userwritten package -estout- by Ben Jann (estout from
http://fmwww.bc.edu/RePEc/bocode/e), and I was able to achieve exactly
what I wanted.

levelsof sample, local(levs)


foreach n in `levs' {
  display ""
  display "================ Regression for level: `n'
============================='"
  logistic outcome var1 var2 var3 if sample==`n', vce(cluster id)
  eststo
  estat ic
}
tabout


Once again, thank you so much for your help.

Best,
Milena

On Tue, Nov 19, 2013 at 8:49 PM, Sergiy Radyakin <serjradyakin@gmail.com> wrote:
> There are many ways to get to the same result. one convenience with
> the loop written explicitly is that it is easier for a beginner to add
> new things into the loop's body, for example, to save results to
> matrices, or dump to a file, and many other tasks. Many Stata commands
> support -by- prefix/suffixes, which can be used to avoid explicit
> looping, at the expense of more constraints at the command.
> Best,Sergiy Radyakin
>
> On Tue, Nov 19, 2013 at 2:40 PM, Nick Winter <njgwinter@gmail.com> wrote:
>> isn't this what the -by:- prefix does:
>>
>> by sample_level, sort: logistic ...
>>
>>
>> - Nick Winter
>>
>>
>>
>> On 11/19/2013 11:11 AM, Sergiy Radyakin wrote:
>>>
>>> Milena, is this something you are after?
>>>
>>> set seed 123
>>> clear
>>> set obs 740
>>> generate sample_level=floor(runiform()*4)
>>> generate outcome=runiform()*4200>2000
>>> generate var1=runiform()
>>> generate var2=runiform()
>>> generate var3=runiform()
>>> generate id=floor(runiform()*2)
>>>
>>> levelsof sample_level, local(levs)
>>>
>>> foreach n in `levs' {
>>>    display ""
>>>    display "======== Regression for level: `n'
>>> ===================================="
>>>    logistic outcome var1 var2 var3 if sample_level==`n', vce(cluster id)
>>> }
>>>
>>> Best, Sergiy
>>>
>>> On Tue, Nov 19, 2013 at 7:34 AM, Milena Przheska
>>> <milena.przheska@cosmicdevelopment.com> wrote:
>>>>
>>>> Dear Statalist members,
>>>>
>>>> I am trying to fit the same logistic regression model using different
>>>> samples from a data set and I am wondering if it is even possible to
>>>> use a -foreach- loop to estimate regression models for each sample
>>>> (indicated by a variable).
>>>>
>>>> I use the variable 'sample' to indicate the sample, but instead of
>>>> writing the command over and over for each sample, I tried to do this
>>>> with a -foreach- loop, something  like this:
>>>>
>>>> egen local sample_level = levelsof(sample)
>>>>
>>>> foreach n of sample_level {
>>>> logistic outcome var1 var2 var3 if sample_level==`n', vce(cluster id)
>>>> }
>>>>
>>>> As always, I would appreciate your help.
>>>>
>>>> Thanks,
>>>> Milena
>>>> *
>>>> *   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/
>>
>>
>> *
>> *   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/
*
*   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/