I would like to contrast two models (one nested within the other) in order
to decide if Model B (the more complex model) provides a significantly
better fit to the data than Model A.
Using the logistic command, it is possible to calculate the marginal
reduction in the -2Log(likelihood), and it is distributed approximately as
Chi-Square. This isn't provided in svylogit because of the complex design.
The logit command allows a pseudo-likelihood to be calculated with a
complex survey design, but my novice reading of Aert et al (2002) "Topic in
modelling clustered data" suggests that pseudo-likelihoods cannot be used
in the calculation of deviance the way likelihoods could be used.
This is a long winded way of asking my question:
How does one contrast nested logistic models estimated with the svylogit
command?
Daniel
*****************
Dr Daniel Reidpath Email: [email protected]
Senior Research Fellow
Social Epidemiology Tel: Local (03) 9244 6717
School of Health and Social Int. +61 3 9244 6717
Development
Deakin University Fax: Local (03) 9244 6017
221 Burwood Hwy Int. +61 3 9244 6017
Burwood VIC 3125
AUSTRALIA
Supported by a VicHealth Senior Research Fellowship
******************
