st: re: total

[Kit Baum <kitbaum@mac.com>]

<>
Martin replied

Your solution amounts to running -total- for each var and
gathering the results in a -matrix-. But what is your take on the  
underlying
question whether a) the default behavior of -total- should be what it  
is and
b) whether there should be a tick box in -db total- which lets me opt  
out of
the current default behavior?


 -total-, like -mean-, -proportion- and -ratio- (which I have not  
tested) are e-class (estimation) commands, not r-class commands like - 
summarize-,  -ameans- and -correlate- (although for the latter - 
pwcorr- is a workaround that addresses this issue, and a not very  
useful one, as it does not leave behind the full r(C)). Although you  
might like to have an option such as you mention above for -total-,  
and it would certainly be feasible, note that most e-class commands  
practice listwise deletion (e.g. -regress-). It is also clear that - 
total- and friends are meant for use in -svy- context, in which I  
imagine anything other than listwise deletion would be dangerous. So  
you can certainly agitate for such an option, but it probably runs  
afoul of established guidelines for the behavior of e-class commands  
to do so.

Kit Baum, Boston College Economics and DIW Berlin
http://ideas.repec.org/e/pba1.html
An Introduction to Modern Econometrics Using Stata:
http://www.stata-press.com/books/imeus.html


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/