On Mon, 1 May 2006, Spencer Graves wrote:
As far as I know, the term deviance has no standard definition. A
good, fairly common definition (I think) is that the deviance is up to
[an additive] constant, minus twice the maximised log-likelihood. Where
sensible, the constant is chosen
Dear Prof. Ripley:
comments in line
Prof Brian Ripley wrote:
On Mon, 1 May 2006, Spencer Graves wrote:
snip
McCullagh Nelder (1989) would be the authorative reference, but the
1982 first edition manages to use 'deviance' in three separate senses on
one page. In particular, what you are
As far as I know, the term deviance has no standard definition. A
good, fairly common definition (I think) is that the deviance is up to
[an additive] constant, minus twice the maximised log-likelihood. Where
sensible, the constant is chosen so that a saturated model has deviance
I often want to know what proportion of the variation is being
contributed by different levels of random effects.
When the random effects are only intercepts, with no slopes, then you
can compute the intra-class correlation, which is the proportion of
the variation explained by the different
Andrew and Spencer,
Thanks for your answers. I was feeling that moving from variance and
least-square estimates to deviance and MLE or REML was the reason why
I could not find easy equivalent estimates for the proportion of
variation of a given effect to the total. Thus this was not a trivial
