Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.

# Re: st: Collecting Statistics of Averages of Variables

 From robert hartman To statalist@hsphsun2.harvard.edu Subject Re: st: Collecting Statistics of Averages of Variables Date Fri, 28 Sep 2012 16:53:31 -0400

```Thanks, Nick. This looks helpful. Where I'm still a bit hung up is the
coding (I assume some kind of loop) to put together all the
combinations, including combinations of varying sizes (e.g.,
3-variable and 2-variable combinations).

I assume there's a fairly straightforward way to do this w/ loops, but
it's not jumping out at me.

Thanks,
Rob

(BTW, failed to mention, I'm on Stata 11.2)

On Fri, Sep 28, 2012 at 2:24 PM, Nick Cox <njcoxstata@gmail.com> wrote:
> Check out -statsby-.
>
> Nick
>
> On 28 Sep 2012, at 18:55, robert hartman <rohartman@gmail.com> wrote:
>
>> Hi Listers,
>> I'm struggling with a couple of programming challenges. I will start
>> with this one. Any input on efficient ways to code this would be most
>> appreciated.
>>
>> I want to take mean (or could be other "egen"/"collapse" type summary
>> statistic) form all possible combinations (of subsets of various
>> sizes) from a set of variables.
>>
>> For example, if I have v1, v2, v3, and v4, I would want the mean of
>> each: v1v2, v1v3, v1v4, v1v2v3, v1v2v4, v1v3v4...v3v2...v3v2v4, etc.
>>
>> By v1v2v3, I mean gen v1v2v3=(v1+v2+v3)/3; (assuming the summary
>> statistic of interest is the mean)
>> by v1v2v3v4, I mean v1v2v3v4=,I mean gen v1v2v3v4=(v1+v2+v3+v4)/4.
>>
>> etc.
>>
>> I don't need to actually create permanent variables, simply
>> 1. create a temporary "combination variable" (e.g., v1v2v4 =
>> rowmean(v1 v2 v4)) for each possible combination,
>> 2. collect summary statistics of interest from "combination variables."
>> 3. spit out a file that conveys this information intelligibly
>>
>> For example, I may want to write out a file that gives (a) the name of
>> the combination (e.g., v1v2v4--the column representing the rowmean of
>> those three variables) and (b) the mean and standard deviation of that
>> v1v2v4 variable.
>>
>> In my particular case, I have about a 70 variable space from which to
>> create all these subset combinations.
>>
>> Any helpful starting points or coding ideas would be helpful.
>>
>> Thanks,
>> Rob Hartman
>> *
>> *   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/
```