Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Nick Cox <njcoxstata@gmail.com> |
To | statalist@hsphsun2.harvard.edu |
Subject | Re: st: How does STATA compute e(ll) ?? |
Date | Thu, 21 Feb 2013 14:59:59 +0000 |
As Maarten underlines separately, this is true in spirit, but not exactly in practice for RMSE. Nick On Thu, Feb 21, 2013 at 2:56 PM, Nick Cox <njcoxstata@gmail.com> wrote: > First off, please note the correct spelling "Stata", as explained in > the Statalist FAQ you were asked to read before posting. > > As you say, the simplest uses of the -regress- command can be thought > of straight applications of ordinary least-squares. But particularly > as a link to more advanced or more general work, Stata reports the > same results as they would be obtained under an equivalent assumption > of maximum likelihood and normal (Gaussian) errors. > > Most of the code for -regress- is part of the executable and not .ado > code. So, my understanding is that -regress- is not built upon any use > of -ml-, but it's safe to assume that key results are exactly > equivalent to that way of writing the command. Somewhere in the > documentation, if I recollect correctly, this is used as an example. > > Nick > > On Thu, Feb 21, 2013 at 2:47 PM, Hsien-Yuan Hsu <hsuhy0914@gmail.com> wrote: > >> When I use commend "regress", I assume OLS regression is conducted. >> Why does STATA report "e(ll) log likelihood under additional >> assumption of i.i.d. normal errors."? Does STATA automatically >> estimate parameter by using ML? > > On Thu, Feb 21, 2013 at 6:08 PM, Nick Cox <njcoxstata@gmail.com> wrote: > >>> In addition, there is a book-length treatment at >>> >>> http://www.stata.com/bookstore/maximum-likelihood-estimation-stata/index.html >>> >>> and Stata's machinery for computing (log) likelihood is documented >>> under heading such as -ml-. > > On Thu, Feb 21, 2013 at 9:21 AM, Nick Cox <njcoxstata@gmail.com> wrote: > >>>> e(ll) is what it says, the log likelihood. Any decent intermediate >>>> statistics text will explain. > > On Thu, Feb 21, 2013 at 8:47 AM, Mark (Hsien-Yuan Hsu) > >>>>> I apply commend "regress Y X" to analyze my data set. >>>>> I notice that STATA provide "e(ll) log likelihood under additional >>>>> assumption of i.i.d. normal errors." >>>>> >>>>> Does anyone know how STATA compute e(ll)? What does it mean? Any >>>>> information is appreciated. * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/faqs/resources/statalist-faq/ * http://www.ats.ucla.edu/stat/stata/