s-news
[Top] [All Lists]

Re: Positive log-likelihood value?

To: dpearce@students.ballarat.edu.au
Subject: Re: Positive log-likelihood value?
From: "Douglas Bates" <bates@stat.wisc.edu>
Date: Sun, 29 Jul 2007 10:06:46 -0500
Cc: s-news@lists.biostat.wustl.edu
Dkim-signature: a=rsa-sha1; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:message-id:date:from:sender:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references:x-google-sender-auth; b=Z8t1xH8f7keuCurqK1sWDPrPmez2BIDpQe647sCoEWNW1G+x7MUgeQ0iL5fuhO7SdDBhFp+8eMnJCcmU67A7chBN3MUYcsVSkJB9sm9EfUrXSMGSYbcHyQjIktjvXG1MTqlXnUMlX8pg+vnc02Ygd+MxEb5A7kF0dpI8622oa0k=
Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:sender:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references:x-google-sender-auth; b=UlT2DZUsVKyr1jm3EdUtCfnD50ZK8YrFxSbbStXosxRIOQTbKoXq0meM2BjMvHi41xGzcL0NI+yUJeEPtX0S/otaPWjUp/FM/um9jZlGKcslOLdCGiJ9A39uLsV5YYlOLK9S0c/HzdxpJDDDHXu8UVrgHzzFqEIRL+uJ5P8gRq0=
In-reply-to: <1185669903.4e00a840dpearce@students.ballarat.edu.au>
References: <1185669903.4e00a840dpearce@students.ballarat.edu.au>
On 7/28/07, Dora C PEARCE <dpearce@students.ballarat.edu.au> wrote:
> Dear S list,
> I have run a spatial linear regression model using a CAR covariance matrix 
> and, in addition to obtaining a significant association between the dependent 
> and independent variables, get a negative log-likelihood (= -6.176). However, 
> when I add a factor and an interaction term as additional explanatory 
> variables, the log-likelihood is positive (= 1.794).
>
> Is it OK to have a POSITIVE log-likelihood, or does this mean there is a 
> misspecification problem?

A positive log-likelihood is acceptable in a model for a continuous
response.  The likelihood in a model for such a response is the
density function evaluated at the observed data and the parameter
values.  A probability density function can be greater than one, which
implies that the log-likelihood can be positive.

<Prev in Thread] Current Thread [Next in Thread>