[RsR] Selecting observations that has smallest absolute residuals

[kh@@x131 m@iii@g oii um@@edu]

Hi Everyone,

I want to pick those observation from the regression data that has smallest 
absolute residuals . i.e i have run ols on 30 observations of multiple 
regression data and got residuals for the whole data. now i want to select 
20 observations from the same data that has smallest absolute residuals and 
want to run ols again on the selected observations. i used the following 
code for selecting 20 observation that has smallest residuals :

 > a<- rank(abs(resid))
 > which(a %in% c(1:20))
 it gives the indices(row numbers) of the regression data that has smallest 
absolute residuals (is it correct?). now how to run ols on this i really 
need your help?

Thanks in advance,
Dost.