RE: Negative Obj Function
Dear Indrajeet,
The objective function value reported is the -2 times Log likelihood and this
is minimized to get the estimates for your model parameters. So when the
log-likelihood is greater than 1, -2* log-likelihood gives a negative value.
Therefore, running in to negative objection value does not affect the precision
of the parameter estimated.
Thanks
Mathangi
Mathangi Gopalakrishnan
Center for Translational Medicine
School of Pharmacy
University of Maryland, Baltimore
Quoted reply history
From: [email protected] [mailto:[email protected]] On
Behalf Of indrajeet singh
Sent: Sunday, May 13, 2012 9:30 PM
To: [email protected]
Subject: [NMusers] Negative Obj Function
What does negative objective function really mean other than the fact that
mathematically speaking likelyhood is greater than 1? Does this affect the
reliability of estimate of parameters? I am running into negative Obj Func but
parameter estimates seems to be ok.
Any comments ?
Best
Jeet
--
Indrajeet Singh,PhD
Sr. Clinical Pharmacokineticist
Abbott Labs, North Chicago, IL