Viridiana,
Suppose you wanted to compute the likelihood that some alternative, b0
is true, and you have estimated the parameter vector b1 (each of length
n). The the LRT requires that you evaluate the log likelihood function
at b0 and at b1. Then 2*(log likelihood of b0 - log likelihood of b1)
is distributed with a chi-square distribution. The degrees of freedom
depends on what you are testing. Suppose you wanted to know if the
first k elements of b1 were really 0, then you could construct b0=b1 and
then replace the first k elements of b0 with 0s. In this example, you
are imposing k restrictions so the test statistic (under the null that
b0 is true) has n-k degrees of freedom.
I hope this is helpful,
Martin
Viridiana Rios wrote:
Hi all,
Does anybody knows the mechanics and R command for applying a
Likelihood-ratio test? I found this at wikipedia
http://en.wikipedia.org/wiki/Likelihood-ratio_test, but I do not have
a clue on what command I should use at R. Any ideas?
Best,
Viridiana R?os
617-997-2471
------------------------------------------------------------------------
_______________________________________________
gov2001-l mailing list
gov2001-l at
lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
--
Martin Sparre Andersen, MPH
Harvard University
Program in Health Policy
anders4 at
fas.harvard.edu
martin.andersen at
aya.yale.edu
AIM: AndersenEPH
Community of Science:
http://myprofile.cos.com/msa25cu