In reading through the digest this morning I noticed that there has been
some discussion
of the AIC statistic. I wrote this little AIC post estimation command a few
months ago that can be used following programs written using Stata's ML
capability. It can be ammended for use with other non-ML regression programs, like
the glm command, by looking at the main program's list of estimation
results. Type
. ereturn list
to see how results have been defined. Then simply change the definitions
listed here as needed. The key is that the main program has to calculate a
log-likelihood function.
Joseph Hilbe
* POST-ESTIMATION CALCULATION OF AIC STATISTIC Joseph Hilbe 7Jul2005
program define aic
version 9.0
tempvar aic
local nobs e(N)
local npred e(df_m)
local df = e(N) - e(df_m) -1
local llike e(ll)
qui gen `aic' = ((-2*`llike') + 2*(`npred'+1))/`nobs'
di in gr _col(1) "AIC Statistic = " in ye %11.3f `aic'
end
*
* 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/
