I of course totally agree with Nick on the general principle.
But if you have something as simple as
xi: impute weekly_hours_of_work i.census_occupation_code, gen(imputed_whw)
and if your data allow, then you want to use (at least) two digits census
occupation code, and that's 84 dummies. In this case, -impute- would
probably be extrely fast (if it weren't limited) because these are 84
dummies.
This is just a "FYI" thread. As far as I am concerned, I am all set.
Best,
Renzo
/* From "Nick Cox" <[email protected]> */
To <[email protected]>
Subject st: RE: Removing the limit to 31 variables from stata -impute- ado
Date Thu, 13 Nov 2003 10:15:07 -0000
[...]
* The performance as the number of predictors
increases. What happens not just to R-sq and
kin but to standard errors etc.? If you can't do
something worthwhile with 31 (carefully
chosen) predictors, it is difficult to believe
that 63 or 127 really would produce a much better
solution. (If they do, you chose the wrong 31.)
* The more general pluses and minuses for what
you are doing, including reproducibility,
avoidance of circular argument and even
avoidance of self-delusion.
Nick
[email protected]
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
