JP writes:
> I'm trying to use estout to format the coefficients of
> several regressions
> as well as the result of an F test that I'm implementing on
> each of of these
> models. I've been trying to use eret2 to store the pvalues of
> my F test, ...
...
> it does not produce the desired effects when I'm trying to store the
> results of multiple models through the command estimates store
> (Example 2).
...
> ********************* Example 2 *********************
>
> sysuse auto, clear
>
> regress price weight mpg
> estimates store m1
> test weight = mpg
> eret2 scalar p_wtmpg = r(p)
>
> regress price weight mpg length turn
> estimates store m2
> test weight = mpg = length = turn
> eret2 scalar p_wtmpg = r(p)
>
> estout m*, stats(N r2 r2_a p_wtmpg) style(fixed)
-eret2- adds stuff to the "active" estimates. However, the
"active" estimates have nothing to do with the "stored"
estimates. Active and stored estimates are two distinct
sets of e()-returns in Stata.
Thus, if you are working with stored estimates and you want
to use -eret2-, always store the estimates AFTER applying
-eret2-. Otherwise, the added statistics will not be part
of the stored estimates and cannot be tabulated by -estout-.
That is, JP's code should be as follows:
. regress price weight mpg
. test weight = mpg
. eret2 scalar p_wtmpg = r(p)
. estimates store m1
. regress price weight mpg length turn
. test weight = mpg = length = turn
. eret2 scalar p_wtmpg = r(p)
. estimates store m2
. estout m*, stats(N r2 r2_a p_wtmpg) style(fixed)
An alternative is to write a routine for -estadd-, that
conducts the F test. Other than -eret2-, -estadd- can
deal with estimates that have already been stored. See
. ssc describe estadd
ben
*
* 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/
