Arnold <[email protected]> asks about -estimates for-:
> I stored results from two separate models and tried without success to test
> equality of parameters across the two result-sets. Here's the sequence:
> svymean varname1
> est store A
> svymean varname2
> est store B
> est for * (or _all) : test [varname1]x = [varname2]y
> For the post-estimation command, I also tried each of the following:
> test A = B
> test [A] = [B]
> test [A=B], common
> In each case, the return said the query to one set of estimates failed in
> the other set of estimates -- which, of course, is true, but I thought from
> the manual that this wouldn't be a program-stopper. I tried with non-survey
> commands (e.g., logit) and got the same results. Help?
There are two issues here:
1. How does -estimates for- work?
The description of -estimates for- in [R] estimates (from the Stata 8 manual
set) says, "... evaluates a post-estimation command for one or more stored
estimation sets." Thus, for example, one could type:
est for res1 res2 res3 : test x1 = x2
instead of typing
tempname curres
qui est store `curres'
foreach res in res1 res2 res3 {
est restore `res'
test x1 = x2
}
est restore `curres'
Thus, testing parameters across estimation results is not a feature of
-estimates for-.
2. How does one go about testing coefficients across different estimation
results?
I would start with [R] suest.
--Jeff
[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/
