Adam wrote:
> I am trying to figure out the easiest way to build a matrix
> of significance stars (or other symbols) after either an
> estimation command or, better yet, after building a table of
> estimates. I figured there would be something like e(stars),
> especially since the estimates table can display them so
> readily. But, short of -parmest-, which makes me build a
> whole new dataset, I'm not sure how to proceed.
It may not be immediately apparent, but -estout- can easily
produce an output table containing only the significance stars.
Here is an example:
. sysuse auto
(1978 Automobile Data)
. reg price mpg foreign
(output suppressed)
. est sto model1
. reg price mpg foreign if uniform()<.4
(output suppressed)
. est sto model2
. estout model1 model2, cells(sham(star)) collabel(,none)
model1 model2
mpg *** *
foreign *
_cons *** ***
In the above example, -cells(sham)- instructs -estout- to report
the parameter statistics stored in e(sham), just as -cells(b)-
would report the contents of e(b). However, since e(sham) does
not exists for the models, empty string is printed.
-collabel(,none)- suppresses printing "sham" above each column.
Note that, you are free to use another name than "sham" in
-cells(...(star))-. For example -cells(sig(star))- or even
-cells(_(star))- would work equally well. The only point is
that e(sham) or whatever does not exist for the models. (There
is one complication tough: the name must not contain "star".)
Possibly, you may also want to mark non-significant coefficients.
This can be done, e.g., as follows:
. estout model1 model2, cells(sham(star)) collabel(,none) ///
> starlevel(n.s. 1 * .05 ** .01 *** .001)
model1 model2
mpg *** *
foreign * n.s.
_cons *** ***
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/