On Tue, 13 Jan 2009, David Winsemius wrote:
I remember have the same consternation using GLIM with binomial models on
grouped and ungrouped data, but I was counseled by my betters only to
consider differences in models. The differences in deviance are the same up
to rounding error.
859.8018
I remember have the same consternation using GLIM with binomial models
on grouped and ungrouped data, but I was counseled by my betters only
to consider differences in models. The differences in deviance are the
same up to rounding error.
> 859.8018 - 711.3479
[1] 148.4539
> 168.8-20.3
[1]
Withut looking up your reference, are you not comparing grouped and
ungrouped deviances? And polr() does not say anything about accepting
a model (or not), only about the comparison between two models.
'Deviances' are in comparison with some 'saturated' model, and I would
say that M&N are com
Dear all,
I've replicated the cheese tasting example on p175 of GLM's by McCullagh
and Nelder. This is a 4 treatment (rows) by 9 ordinal response (cols)
table.
Here's my simple code:
cheese
library(MASS)
options(contrasts = c("contr.treatment", "contr.poly"))
y = c
4 matches
Mail list logo
