st: RE: AW: RE: AW: RE: AW: Exclude some observations from a sample in a regression

["Nick Cox" <n.j.cox@durham.ac.uk>]

Good catch. 

Nick 
n.j.cox@durham.ac.uk 

Martin Weiss

" You may intend 

... if foreign == "foreign":origin"

This will lead to 

no observations
r(2000);

You must capitalize "Foreign" for this to work:

*************
sysuse auto, clear
regress price weight length if foreign == "Foreign":origin
regress price weight length if foreign == "foreign":origin
*************

Nick Cox

The overarching principle is that 

... if <condition> 

identifies the maximal subset that will be included. That doesn't stop
observations being excluded on other grounds, namely missing values of
-rep78- in these examples. 

Note that 

... if foreign == foreign 

is no exclusion at all, being on all fours with -if 1 == 1- or -if 42 ==
42-, that is, it is a tautology, or always true. You may intend 

... if foreign == "foreign":origin 

which is quite different. See

SJ-4-4  dm0009  . . . . . . .  Stata tip 14: Using value labels in
expressions
        . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  K.
Higbee
        Q4/04   SJ 4(4):488--489                                 (no
commands)
        tips for using value labels in expressions

Nick 
n.j.cox@durham.ac.uk 

Claude Francoeur

Thanks for showing me these different methods. I should have mentioned 
that I do have missing values in my sample. I'm confused though 
concerning the right command to use.

regress price weight length if foreign returns a regression using 22 
observations since there are no missing cases

regress price weight length rep78 if foreign returns a regression using 
21 observations since there is one missing value under rep78. This seems

to be contrary to Nick's comment, but I probably am missing something. 


regress price weight length rep78 if foreign==foreign returns a 
regression using all 69 non missing observations including domestic 
cases. The if command does not seem to be recognised.

*
*   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/