Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Shourjo Chakravorty <shourjo@gmail.com> |
To | statalist@hsphsun2.harvard.edu |
Subject | Re: st: Backed-up message on last iteration |
Date | Tue, 4 Oct 2011 11:40:48 -0400 |
Hi Austin, The method you suggested is exactly what I have been doing. I still get the backed-up message on the last iteration when using the estimated coef vector as the starting point. Although in this case I have the options "partial", "technique()" and "difficult" on. Should I turn them off and do the biprobit estimation with the estimated coefficient vector as the starting point? And also if these results cannot be trusted, should I be checking the gradient vector? Shourjo On 4 October 2011 10:18, Austin Nichols <austinnichols@gmail.com> wrote: > Shourjo Chakravorty <shourjo@gmail.com> : > > What happens when you restart -biprobit- without the options, using > the estimated coef vector as your starting point? > > webuse school, clear > biprobit private vote logptax loginc years, tech(bfgs) > mat b=e(b) > biprobit private vote logptax loginc years, from(b) > > Only you can clarify your confusion. > > On Mon, Oct 3, 2011 at 10:19 PM, Shourjo Chakravorty <shourjo@gmail.com> wrote: >> Btw I am using Stata11. >> >> On 3 October 2011 22:02, Shourjo Chakravorty <shourjo@gmail.com> wrote: >>> Hello, >>> >>> I am estimating a partially-observable biprobit model with the dfp and >>> bfgs techniques.Sometimes I get the "backed-up message" on the last >>> iteration. I've read the archives here and it says I can trust the >>> results. I am asking because in the book by Gould, Pitblando and Poi >>> (2010) "maximum likelihood estimation with stata" it states that I >>> cannot trust the results if the last iteration has this message (page >>> 20). I am a beginner and would be grateful if someone could clarify my >>> confusion. >>> > * > * For searches and help try: > * http://www.stata.com/help.cgi?search > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ > * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/