Re: st: Unexpected output from -collapse (rawsum)

[khigbee@stata.com]

James Beard <james@beard.net> demonstrated that -collapse- with
the -rawsum- statistic does not include in the sum any
observations that have zero weight.  His code was:

> . clear all
> . set obs 2
> obs was 0, now 2
> . gen pweight = _n - 1
> . gen v = 1
> . list, clean
>        pweight   v  
>   1.         0   1  
>   2.         1   1  
> . collapse (rawsum) rs=v (sum) ws=v [pw=pweight]
> . list, clean
>        rs   ws  
>   1.    1    1  

And James commented:

> I would have expected rs to be 2 (that is, the same as the result of 
> unweighted -collapse (sum)-. What appears to be happening is that 
> only records with 0<pweight<. are included in the output.
> The documentation, however, says that -rawsum- provides "sums,
> ignoring optionally specified weight".

In -collapse-, zero weighted observations are removed from the
computations (just like we do for observations excluded based on
-if- or -in-).  This is done before -rawsum- (and all the other
available -collapse- statistics) are calculated.  We will modify
the documentation to clarify this point.


Ken Higbee    khigbee@stata.com
StataCorp     1-800-STATAPC

*
*   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/