[R] Bug or Feature? LogLik.nls and non-central F distribution.
Matthew L. Fidler
fidler at math.utah.edu
Wed Jan 15 17:38:38 CET 2003
On Wed, 2003-01-15 at 09:18, Douglas Bates wrote:
> "Matthew L. Fidler" <fidler at math.utah.edu> writes:
>
> ...
> > > logLik(short.alif.fit);
> > `log Lik.' 6.307024 (df=1)
> >
> > My problem with this number is that exp(logLik) > 1. If the error
> > structure is independent, identically distributed normal (which I
> > assumed was the case), then the Likelihood function should only give a
> > number from 0 to 1...
>
> Why? The likelihood is the product of the probability densities of
> the observations given the parameters and a probability density can be
> greater than 1.
>
> The belief that a likelihood must be less than 1 is a common
> misconception
True, if the variance is small (less than one), it can have that
feature. I forgot; thank you for reminding me. However I still cannot
get the non-central F to return the right tail (according to my linear
model's noncentral F distribution tables....) I thought it wasn't
working because I couldn't get it to produce the log-liklihood function
that I derived. I forgot to divide the residual sum of squares by N,
however.... sorry.
--
Matthew L. Fidler
fidler at math.utah.edu
421 Wakra Way, Suite 318
Salt Lake City, UT 84108
(801) 581-7125
-------------------------------------------------------------------------
Accident, n.:
A condition in which presence of mind is good, but absence of
body is better.
-- Foolish Dictionary
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : https://stat.ethz.ch/pipermail/r-help/attachments/20030115/6b1bebb0/attachment.bin
More information about the R-help
mailing list