Unless memory is limited, you can avoid confusions like this by not
using -drop- at all. Just let Stata determine what to omit.
Nick
[email protected]
Francesco Trivieri
thank you very much. I eventually understood the problem. It was
depending
on the lagged dependent variable. If I do
1) drop if x>123
rather than
2) drop if x>123 & x!=.
I am eliminating observations that Stata may use to calculate Y-1. So, I
avoid the problem by using 2).
From: "Martin Weiss" <[email protected]>
> Well, in the absence of an example that is reproducible to all it is
> difficult to diagnose the problem. In my example, both approaches end
up
> delivering the same nobs...
>
>
> *************
> use http://www.stata-press.com/data/r7/abdata.dta, clear
>
> replace w=. in 200/250
>
> preserve
> drop if w>3.4
> xtabond2 n w k, iv(w k, eq(level)) small h(1)
> restore
>
> preserve
> drop if w>3.4 & w!=.
> xtabond2 n w k, iv(w k, eq(level)) small h(1)
> restore
> *************
>
Francesco Trivieri
> thank you very much for your answers. I am always including the x
variable
> in my estimation, I don't understand why:
>
> drop if x>123
> xtabond2 Y Y_1 X Z ....
> the estimation sample is 8100
>
> drop if x>123 & x!=.
> xtabond2 Y Y_1 X Z ....
> the estimation sample is 12200
*
* 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/
